linux/sound/pci/hda/hda_intel.c

2692 lines
75 KiB
C
Raw Normal View History

/*
*
* hda_intel.c - Implementation of primary alsa driver code base
* for Intel HD Audio.
*
* Copyright(c) 2004 Intel Corporation. All rights reserved.
*
* Copyright (c) 2004 Takashi Iwai <tiwai@suse.de>
* PeiSen Hou <pshou@realtek.com.tw>
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the Free
* Software Foundation; either version 2 of the License, or (at your option)
* any later version.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License along with
* this program; if not, write to the Free Software Foundation, Inc., 59
* Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
* CONTACTS:
*
* Matt Jared matt.jared@intel.com
* Andy Kopp andy.kopp@intel.com
* Dan Kogan dan.d.kogan@intel.com
*
* CHANGES:
*
* 2004.12.01 Major rewrite by tiwai, merged the work of pshou
*
*/
#include <linux/delay.h>
#include <linux/interrupt.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/dma-mapping.h>
#include <linux/moduleparam.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/pci.h>
#include <linux/mutex.h>
#include <linux/io.h>
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
#include <linux/pm_runtime.h>
#include <linux/clocksource.h>
#include <linux/time.h>
#include <linux/completion.h>
#ifdef CONFIG_X86
/* for snoop control */
#include <asm/pgtable.h>
#include <asm/set_memory.h>
#include <asm/cpufeature.h>
#endif
#include <sound/core.h>
#include <sound/initval.h>
#include <sound/hdaudio.h>
#include <sound/hda_i915.h>
#include <linux/vgaarb.h>
#include <linux/vga_switcheroo.h>
#include <linux/firmware.h>
#include <sound/hda_codec.h>
#include "hda_controller.h"
#include "hda_intel.h"
#define CREATE_TRACE_POINTS
#include "hda_intel_trace.h"
/* position fix mode */
enum {
POS_FIX_AUTO,
POS_FIX_LPIB,
POS_FIX_POSBUF,
POS_FIX_VIACOMBO,
POS_FIX_COMBO,
POS_FIX_SKL,
};
/* Defines for ATI HD Audio support in SB450 south bridge */
#define ATI_SB450_HDAUDIO_MISC_CNTR2_ADDR 0x42
#define ATI_SB450_HDAUDIO_ENABLE_SNOOP 0x02
/* Defines for Nvidia HDA support */
#define NVIDIA_HDA_TRANSREG_ADDR 0x4e
#define NVIDIA_HDA_ENABLE_COHBITS 0x0f
#define NVIDIA_HDA_ISTRM_COH 0x4d
#define NVIDIA_HDA_OSTRM_COH 0x4c
#define NVIDIA_HDA_ENABLE_COHBIT 0x01
/* Defines for Intel SCH HDA snoop control */
#define INTEL_HDA_CGCTL 0x48
#define INTEL_HDA_CGCTL_MISCBDCGE (0x1 << 6)
#define INTEL_SCH_HDA_DEVC 0x78
#define INTEL_SCH_HDA_DEVC_NOSNOOP (0x1<<11)
/* Define IN stream 0 FIFO size offset in VIA controller */
#define VIA_IN_STREAM0_FIFO_SIZE_OFFSET 0x90
/* Define VIA HD Audio Device ID*/
#define VIA_HDAC_DEVICE_ID 0x3288
/* max number of SDs */
/* ICH, ATI and VIA have 4 playback and 4 capture */
#define ICH6_NUM_CAPTURE 4
#define ICH6_NUM_PLAYBACK 4
/* ULI has 6 playback and 5 capture */
#define ULI_NUM_CAPTURE 5
#define ULI_NUM_PLAYBACK 6
/* ATI HDMI may have up to 8 playbacks and 0 capture */
#define ATIHDMI_NUM_CAPTURE 0
#define ATIHDMI_NUM_PLAYBACK 8
/* TERA has 4 playback and 3 capture */
#define TERA_NUM_CAPTURE 3
#define TERA_NUM_PLAYBACK 4
static int index[SNDRV_CARDS] = SNDRV_DEFAULT_IDX;
static char *id[SNDRV_CARDS] = SNDRV_DEFAULT_STR;
static bool enable[SNDRV_CARDS] = SNDRV_DEFAULT_ENABLE_PNP;
static char *model[SNDRV_CARDS];
static int position_fix[SNDRV_CARDS] = {[0 ... (SNDRV_CARDS-1)] = -1};
static int bdl_pos_adj[SNDRV_CARDS] = {[0 ... (SNDRV_CARDS-1)] = -1};
static int probe_mask[SNDRV_CARDS] = {[0 ... (SNDRV_CARDS-1)] = -1};
static int probe_only[SNDRV_CARDS];
static int jackpoll_ms[SNDRV_CARDS];
static int single_cmd = -1;
static int enable_msi = -1;
#ifdef CONFIG_SND_HDA_PATCH_LOADER
static char *patch[SNDRV_CARDS];
#endif
#ifdef CONFIG_SND_HDA_INPUT_BEEP
static bool beep_mode[SNDRV_CARDS] = {[0 ... (SNDRV_CARDS-1)] =
CONFIG_SND_HDA_INPUT_BEEP_MODE};
#endif
module_param_array(index, int, NULL, 0444);
MODULE_PARM_DESC(index, "Index value for Intel HD audio interface.");
module_param_array(id, charp, NULL, 0444);
MODULE_PARM_DESC(id, "ID string for Intel HD audio interface.");
module_param_array(enable, bool, NULL, 0444);
MODULE_PARM_DESC(enable, "Enable Intel HD audio interface.");
module_param_array(model, charp, NULL, 0444);
MODULE_PARM_DESC(model, "Use the given board model.");
module_param_array(position_fix, int, NULL, 0444);
MODULE_PARM_DESC(position_fix, "DMA pointer read method."
"(-1 = system default, 0 = auto, 1 = LPIB, 2 = POSBUF, 3 = VIACOMBO, 4 = COMBO, 5 = SKL+).");
module_param_array(bdl_pos_adj, int, NULL, 0644);
MODULE_PARM_DESC(bdl_pos_adj, "BDL position adjustment offset.");
module_param_array(probe_mask, int, NULL, 0444);
MODULE_PARM_DESC(probe_mask, "Bitmask to probe codecs (default = -1).");
module_param_array(probe_only, int, NULL, 0444);
MODULE_PARM_DESC(probe_only, "Only probing and no codec initialization.");
module_param_array(jackpoll_ms, int, NULL, 0444);
MODULE_PARM_DESC(jackpoll_ms, "Ms between polling for jack events (default = 0, using unsol events only)");
module_param(single_cmd, bint, 0444);
MODULE_PARM_DESC(single_cmd, "Use single command to communicate with codecs "
"(for debugging only).");
module_param(enable_msi, bint, 0444);
MODULE_PARM_DESC(enable_msi, "Enable Message Signaled Interrupt (MSI)");
#ifdef CONFIG_SND_HDA_PATCH_LOADER
module_param_array(patch, charp, NULL, 0444);
MODULE_PARM_DESC(patch, "Patch file for Intel HD audio interface.");
#endif
#ifdef CONFIG_SND_HDA_INPUT_BEEP
module_param_array(beep_mode, bool, NULL, 0444);
MODULE_PARM_DESC(beep_mode, "Select HDA Beep registration mode "
"(0=off, 1=on) (default=1).");
#endif
#ifdef CONFIG_PM
static int param_set_xint(const char *val, const struct kernel_param *kp);
static const struct kernel_param_ops param_ops_xint = {
.set = param_set_xint,
.get = param_get_int,
};
#define param_check_xint param_check_int
static int power_save = CONFIG_SND_HDA_POWER_SAVE_DEFAULT;
module_param(power_save, xint, 0644);
MODULE_PARM_DESC(power_save, "Automatic power-saving timeout "
"(in second, 0 = disable).");
static bool pm_blacklist = true;
module_param(pm_blacklist, bool, 0644);
MODULE_PARM_DESC(pm_blacklist, "Enable power-management blacklist");
/* reset the HD-audio controller in power save mode.
* this may give more power-saving, but will take longer time to
* wake up.
*/
static bool power_save_controller = 1;
module_param(power_save_controller, bool, 0644);
MODULE_PARM_DESC(power_save_controller, "Reset controller in power save mode.");
#else
#define power_save 0
#endif /* CONFIG_PM */
static int align_buffer_size = -1;
module_param(align_buffer_size, bint, 0644);
MODULE_PARM_DESC(align_buffer_size,
"Force buffer and period sizes to be multiple of 128 bytes.");
#ifdef CONFIG_X86
static int hda_snoop = -1;
module_param_named(snoop, hda_snoop, bint, 0444);
MODULE_PARM_DESC(snoop, "Enable/disable snooping");
#else
#define hda_snoop true
#endif
MODULE_LICENSE("GPL");
MODULE_SUPPORTED_DEVICE("{{Intel, ICH6},"
"{Intel, ICH6M},"
"{Intel, ICH7},"
"{Intel, ESB2},"
"{Intel, ICH8},"
"{Intel, ICH9},"
"{Intel, ICH10},"
"{Intel, PCH},"
"{Intel, CPT},"
"{Intel, PPT},"
"{Intel, LPT},"
"{Intel, LPT_LP},"
"{Intel, WPT_LP},"
"{Intel, SPT},"
"{Intel, SPT_LP},"
"{Intel, HPT},"
"{Intel, PBG},"
"{Intel, SCH},"
"{ATI, SB450},"
"{ATI, SB600},"
"{ATI, RS600},"
"{ATI, RS690},"
"{ATI, RS780},"
"{ATI, R600},"
"{ATI, RV630},"
"{ATI, RV610},"
"{ATI, RV670},"
"{ATI, RV635},"
"{ATI, RV620},"
"{ATI, RV770},"
"{VIA, VT8251},"
"{VIA, VT8237A},"
"{SiS, SIS966},"
"{ULI, M5461}}");
MODULE_DESCRIPTION("Intel HDA driver");
#if defined(CONFIG_PM) && defined(CONFIG_VGA_SWITCHEROO)
#if IS_ENABLED(CONFIG_SND_HDA_CODEC_HDMI)
#define SUPPORT_VGA_SWITCHEROO
#endif
#endif
/*
*/
/* driver types */
enum {
AZX_DRIVER_ICH,
AZX_DRIVER_PCH,
AZX_DRIVER_SCH,
AZX_DRIVER_SKL,
AZX_DRIVER_HDMI,
AZX_DRIVER_ATI,
AZX_DRIVER_ATIHDMI,
AZX_DRIVER_ATIHDMI_NS,
AZX_DRIVER_VIA,
AZX_DRIVER_SIS,
AZX_DRIVER_ULI,
AZX_DRIVER_NVIDIA,
AZX_DRIVER_TERA,
AZX_DRIVER_CTX,
AZX_DRIVER_CTHDA,
AZX_DRIVER_CMEDIA,
AZX_DRIVER_GENERIC,
AZX_NUM_DRIVERS, /* keep this as last entry */
};
#define azx_get_snoop_type(chip) \
(((chip)->driver_caps & AZX_DCAPS_SNOOP_MASK) >> 10)
#define AZX_DCAPS_SNOOP_TYPE(type) ((AZX_SNOOP_TYPE_ ## type) << 10)
/* quirks for old Intel chipsets */
#define AZX_DCAPS_INTEL_ICH \
(AZX_DCAPS_OLD_SSYNC | AZX_DCAPS_NO_ALIGN_BUFSIZE)
/* quirks for Intel PCH */
#define AZX_DCAPS_INTEL_PCH_BASE \
(AZX_DCAPS_NO_ALIGN_BUFSIZE | AZX_DCAPS_COUNT_LPIB_DELAY |\
AZX_DCAPS_SNOOP_TYPE(SCH))
/* PCH up to IVB; no runtime PM; bind with i915 gfx */
#define AZX_DCAPS_INTEL_PCH_NOPM \
(AZX_DCAPS_INTEL_PCH_BASE | AZX_DCAPS_I915_COMPONENT)
/* PCH for HSW/BDW; with runtime PM */
/* no i915 binding for this as HSW/BDW has another controller for HDMI */
#define AZX_DCAPS_INTEL_PCH \
(AZX_DCAPS_INTEL_PCH_BASE | AZX_DCAPS_PM_RUNTIME)
/* HSW HDMI */
#define AZX_DCAPS_INTEL_HASWELL \
(/*AZX_DCAPS_ALIGN_BUFSIZE |*/ AZX_DCAPS_COUNT_LPIB_DELAY |\
AZX_DCAPS_PM_RUNTIME | AZX_DCAPS_I915_COMPONENT |\
AZX_DCAPS_I915_POWERWELL | AZX_DCAPS_SNOOP_TYPE(SCH))
/* Broadwell HDMI can't use position buffer reliably, force to use LPIB */
#define AZX_DCAPS_INTEL_BROADWELL \
(/*AZX_DCAPS_ALIGN_BUFSIZE |*/ AZX_DCAPS_POSFIX_LPIB |\
AZX_DCAPS_PM_RUNTIME | AZX_DCAPS_I915_COMPONENT |\
AZX_DCAPS_I915_POWERWELL | AZX_DCAPS_SNOOP_TYPE(SCH))
#define AZX_DCAPS_INTEL_BAYTRAIL \
(AZX_DCAPS_INTEL_PCH_BASE | AZX_DCAPS_I915_COMPONENT |\
AZX_DCAPS_I915_POWERWELL)
#define AZX_DCAPS_INTEL_BRASWELL \
(AZX_DCAPS_INTEL_PCH_BASE | AZX_DCAPS_PM_RUNTIME |\
AZX_DCAPS_I915_COMPONENT | AZX_DCAPS_I915_POWERWELL)
#define AZX_DCAPS_INTEL_SKYLAKE \
(AZX_DCAPS_INTEL_PCH_BASE | AZX_DCAPS_PM_RUNTIME |\
AZX_DCAPS_SEPARATE_STREAM_TAG | AZX_DCAPS_I915_COMPONENT |\
AZX_DCAPS_I915_POWERWELL)
#define AZX_DCAPS_INTEL_BROXTON \
(AZX_DCAPS_INTEL_PCH_BASE | AZX_DCAPS_PM_RUNTIME |\
AZX_DCAPS_SEPARATE_STREAM_TAG | AZX_DCAPS_I915_COMPONENT |\
AZX_DCAPS_I915_POWERWELL)
/* quirks for ATI SB / AMD Hudson */
#define AZX_DCAPS_PRESET_ATI_SB \
(AZX_DCAPS_NO_TCSEL | AZX_DCAPS_SYNC_WRITE | AZX_DCAPS_POSFIX_LPIB |\
AZX_DCAPS_SNOOP_TYPE(ATI))
/* quirks for ATI/AMD HDMI */
#define AZX_DCAPS_PRESET_ATI_HDMI \
(AZX_DCAPS_NO_TCSEL | AZX_DCAPS_SYNC_WRITE | AZX_DCAPS_POSFIX_LPIB|\
AZX_DCAPS_NO_MSI64)
/* quirks for ATI HDMI with snoop off */
#define AZX_DCAPS_PRESET_ATI_HDMI_NS \
(AZX_DCAPS_PRESET_ATI_HDMI | AZX_DCAPS_SNOOP_OFF)
/* quirks for Nvidia */
#define AZX_DCAPS_PRESET_NVIDIA \
(AZX_DCAPS_NO_MSI | AZX_DCAPS_CORBRP_SELF_CLEAR |\
AZX_DCAPS_SNOOP_TYPE(NVIDIA))
#define AZX_DCAPS_PRESET_CTHDA \
(AZX_DCAPS_NO_MSI | AZX_DCAPS_POSFIX_LPIB |\
AZX_DCAPS_NO_64BIT |\
AZX_DCAPS_4K_BDLE_BOUNDARY | AZX_DCAPS_SNOOP_OFF)
/*
* vga_switcheroo support
*/
#ifdef SUPPORT_VGA_SWITCHEROO
#define use_vga_switcheroo(chip) ((chip)->use_vga_switcheroo)
#else
#define use_vga_switcheroo(chip) 0
#endif
#define CONTROLLER_IN_GPU(pci) (((pci)->device == 0x0a0c) || \
((pci)->device == 0x0c0c) || \
((pci)->device == 0x0d0c) || \
((pci)->device == 0x160c))
#define IS_BXT(pci) ((pci)->vendor == 0x8086 && (pci)->device == 0x5a98)
#define IS_CFL(pci) ((pci)->vendor == 0x8086 && (pci)->device == 0xa348)
static char *driver_short_names[] = {
[AZX_DRIVER_ICH] = "HDA Intel",
[AZX_DRIVER_PCH] = "HDA Intel PCH",
[AZX_DRIVER_SCH] = "HDA Intel MID",
[AZX_DRIVER_SKL] = "HDA Intel PCH", /* kept old name for compatibility */
[AZX_DRIVER_HDMI] = "HDA Intel HDMI",
[AZX_DRIVER_ATI] = "HDA ATI SB",
[AZX_DRIVER_ATIHDMI] = "HDA ATI HDMI",
[AZX_DRIVER_ATIHDMI_NS] = "HDA ATI HDMI",
[AZX_DRIVER_VIA] = "HDA VIA VT82xx",
[AZX_DRIVER_SIS] = "HDA SIS966",
[AZX_DRIVER_ULI] = "HDA ULI M5461",
[AZX_DRIVER_NVIDIA] = "HDA NVidia",
[AZX_DRIVER_TERA] = "HDA Teradici",
[AZX_DRIVER_CTX] = "HDA Creative",
[AZX_DRIVER_CTHDA] = "HDA Creative",
[AZX_DRIVER_CMEDIA] = "HDA C-Media",
[AZX_DRIVER_GENERIC] = "HD-Audio Generic",
};
#ifdef CONFIG_X86
static void __mark_pages_wc(struct azx *chip, struct snd_dma_buffer *dmab, bool on)
{
int pages;
if (azx_snoop(chip))
return;
if (!dmab || !dmab->area || !dmab->bytes)
return;
#ifdef CONFIG_SND_DMA_SGBUF
if (dmab->dev.type == SNDRV_DMA_TYPE_DEV_SG) {
struct snd_sg_buf *sgbuf = dmab->private_data;
if (chip->driver_type == AZX_DRIVER_CMEDIA)
return; /* deal with only CORB/RIRB buffers */
if (on)
set_pages_array_wc(sgbuf->page_table, sgbuf->pages);
else
set_pages_array_wb(sgbuf->page_table, sgbuf->pages);
return;
}
#endif
pages = (dmab->bytes + PAGE_SIZE - 1) >> PAGE_SHIFT;
if (on)
set_memory_wc((unsigned long)dmab->area, pages);
else
set_memory_wb((unsigned long)dmab->area, pages);
}
static inline void mark_pages_wc(struct azx *chip, struct snd_dma_buffer *buf,
bool on)
{
__mark_pages_wc(chip, buf, on);
}
static inline void mark_runtime_wc(struct azx *chip, struct azx_dev *azx_dev,
struct snd_pcm_substream *substream, bool on)
{
if (azx_dev->wc_marked != on) {
__mark_pages_wc(chip, snd_pcm_get_dma_buf(substream), on);
azx_dev->wc_marked = on;
}
}
#else
/* NOP for other archs */
static inline void mark_pages_wc(struct azx *chip, struct snd_dma_buffer *buf,
bool on)
{
}
static inline void mark_runtime_wc(struct azx *chip, struct azx_dev *azx_dev,
struct snd_pcm_substream *substream, bool on)
{
}
#endif
static int azx_acquire_irq(struct azx *chip, int do_disconnect);
/*
* initialize the PCI registers
*/
/* update bits in a PCI register byte */
static void update_pci_byte(struct pci_dev *pci, unsigned int reg,
unsigned char mask, unsigned char val)
{
unsigned char data;
pci_read_config_byte(pci, reg, &data);
data &= ~mask;
data |= (val & mask);
pci_write_config_byte(pci, reg, data);
}
static void azx_init_pci(struct azx *chip)
{
int snoop_type = azx_get_snoop_type(chip);
/* Clear bits 0-2 of PCI register TCSEL (at offset 0x44)
* TCSEL == Traffic Class Select Register, which sets PCI express QOS
* Ensuring these bits are 0 clears playback static on some HD Audio
* codecs.
* The PCI register TCSEL is defined in the Intel manuals.
*/
if (!(chip->driver_caps & AZX_DCAPS_NO_TCSEL)) {
dev_dbg(chip->card->dev, "Clearing TCSEL\n");
update_pci_byte(chip->pci, AZX_PCIREG_TCSEL, 0x07, 0);
}
/* For ATI SB450/600/700/800/900 and AMD Hudson azalia HD audio,
* we need to enable snoop.
*/
if (snoop_type == AZX_SNOOP_TYPE_ATI) {
dev_dbg(chip->card->dev, "Setting ATI snoop: %d\n",
azx_snoop(chip));
update_pci_byte(chip->pci,
ATI_SB450_HDAUDIO_MISC_CNTR2_ADDR, 0x07,
azx_snoop(chip) ? ATI_SB450_HDAUDIO_ENABLE_SNOOP : 0);
}
/* For NVIDIA HDA, enable snoop */
if (snoop_type == AZX_SNOOP_TYPE_NVIDIA) {
dev_dbg(chip->card->dev, "Setting Nvidia snoop: %d\n",
azx_snoop(chip));
update_pci_byte(chip->pci,
NVIDIA_HDA_TRANSREG_ADDR,
0x0f, NVIDIA_HDA_ENABLE_COHBITS);
update_pci_byte(chip->pci,
NVIDIA_HDA_ISTRM_COH,
0x01, NVIDIA_HDA_ENABLE_COHBIT);
update_pci_byte(chip->pci,
NVIDIA_HDA_OSTRM_COH,
0x01, NVIDIA_HDA_ENABLE_COHBIT);
}
/* Enable SCH/PCH snoop if needed */
if (snoop_type == AZX_SNOOP_TYPE_SCH) {
unsigned short snoop;
pci_read_config_word(chip->pci, INTEL_SCH_HDA_DEVC, &snoop);
if ((!azx_snoop(chip) && !(snoop & INTEL_SCH_HDA_DEVC_NOSNOOP)) ||
(azx_snoop(chip) && (snoop & INTEL_SCH_HDA_DEVC_NOSNOOP))) {
snoop &= ~INTEL_SCH_HDA_DEVC_NOSNOOP;
if (!azx_snoop(chip))
snoop |= INTEL_SCH_HDA_DEVC_NOSNOOP;
pci_write_config_word(chip->pci, INTEL_SCH_HDA_DEVC, snoop);
pci_read_config_word(chip->pci,
INTEL_SCH_HDA_DEVC, &snoop);
}
dev_dbg(chip->card->dev, "SCH snoop: %s\n",
(snoop & INTEL_SCH_HDA_DEVC_NOSNOOP) ?
"Disabled" : "Enabled");
}
}
/*
* In BXT-P A0, HD-Audio DMA requests is later than expected,
* and makes an audio stream sensitive to system latencies when
* 24/32 bits are playing.
* Adjusting threshold of DMA fifo to force the DMA request
* sooner to improve latency tolerance at the expense of power.
*/
static void bxt_reduce_dma_latency(struct azx *chip)
{
u32 val;
val = azx_readl(chip, VS_EM4L);
val &= (0x3 << 20);
azx_writel(chip, VS_EM4L, val);
}
/*
* ML_LCAP bits:
* bit 0: 6 MHz Supported
* bit 1: 12 MHz Supported
* bit 2: 24 MHz Supported
* bit 3: 48 MHz Supported
* bit 4: 96 MHz Supported
* bit 5: 192 MHz Supported
*/
static int intel_get_lctl_scf(struct azx *chip)
{
struct hdac_bus *bus = azx_bus(chip);
static int preferred_bits[] = { 2, 3, 1, 4, 5 };
u32 val, t;
int i;
val = readl(bus->mlcap + AZX_ML_BASE + AZX_REG_ML_LCAP);
for (i = 0; i < ARRAY_SIZE(preferred_bits); i++) {
t = preferred_bits[i];
if (val & (1 << t))
return t;
}
dev_warn(chip->card->dev, "set audio clock frequency to 6MHz");
return 0;
}
static int intel_ml_lctl_set_power(struct azx *chip, int state)
{
struct hdac_bus *bus = azx_bus(chip);
u32 val;
int timeout;
/*
* the codecs are sharing the first link setting by default
* If other links are enabled for stream, they need similar fix
*/
val = readl(bus->mlcap + AZX_ML_BASE + AZX_REG_ML_LCTL);
val &= ~AZX_MLCTL_SPA;
val |= state << AZX_MLCTL_SPA_SHIFT;
writel(val, bus->mlcap + AZX_ML_BASE + AZX_REG_ML_LCTL);
/* wait for CPA */
timeout = 50;
while (timeout) {
if (((readl(bus->mlcap + AZX_ML_BASE + AZX_REG_ML_LCTL)) &
AZX_MLCTL_CPA) == (state << AZX_MLCTL_CPA_SHIFT))
return 0;
timeout--;
udelay(10);
}
return -1;
}
static void intel_init_lctl(struct azx *chip)
{
struct hdac_bus *bus = azx_bus(chip);
u32 val;
int ret;
/* 0. check lctl register value is correct or not */
val = readl(bus->mlcap + AZX_ML_BASE + AZX_REG_ML_LCTL);
/* if SCF is already set, let's use it */
if ((val & ML_LCTL_SCF_MASK) != 0)
return;
/*
* Before operating on SPA, CPA must match SPA.
* Any deviation may result in undefined behavior.
*/
if (((val & AZX_MLCTL_SPA) >> AZX_MLCTL_SPA_SHIFT) !=
((val & AZX_MLCTL_CPA) >> AZX_MLCTL_CPA_SHIFT))
return;
/* 1. turn link down: set SPA to 0 and wait CPA to 0 */
ret = intel_ml_lctl_set_power(chip, 0);
udelay(100);
if (ret)
goto set_spa;
/* 2. update SCF to select a properly audio clock*/
val &= ~ML_LCTL_SCF_MASK;
val |= intel_get_lctl_scf(chip);
writel(val, bus->mlcap + AZX_ML_BASE + AZX_REG_ML_LCTL);
set_spa:
/* 4. turn link up: set SPA to 1 and wait CPA to 1 */
intel_ml_lctl_set_power(chip, 1);
udelay(100);
}
static void hda_intel_init_chip(struct azx *chip, bool full_reset)
{
struct hdac_bus *bus = azx_bus(chip);
struct pci_dev *pci = chip->pci;
u32 val;
if (chip->driver_caps & AZX_DCAPS_I915_POWERWELL)
snd_hdac_set_codec_wakeup(bus, true);
if (chip->driver_type == AZX_DRIVER_SKL) {
pci_read_config_dword(pci, INTEL_HDA_CGCTL, &val);
val = val & ~INTEL_HDA_CGCTL_MISCBDCGE;
pci_write_config_dword(pci, INTEL_HDA_CGCTL, val);
}
azx_init_chip(chip, full_reset);
if (chip->driver_type == AZX_DRIVER_SKL) {
pci_read_config_dword(pci, INTEL_HDA_CGCTL, &val);
val = val | INTEL_HDA_CGCTL_MISCBDCGE;
pci_write_config_dword(pci, INTEL_HDA_CGCTL, val);
}
if (chip->driver_caps & AZX_DCAPS_I915_POWERWELL)
snd_hdac_set_codec_wakeup(bus, false);
/* reduce dma latency to avoid noise */
if (IS_BXT(pci))
bxt_reduce_dma_latency(chip);
if (bus->mlcap != NULL)
intel_init_lctl(chip);
}
/* calculate runtime delay from LPIB */
static int azx_get_delay_from_lpib(struct azx *chip, struct azx_dev *azx_dev,
unsigned int pos)
{
struct snd_pcm_substream *substream = azx_dev->core.substream;
int stream = substream->stream;
unsigned int lpib_pos = azx_get_pos_lpib(chip, azx_dev);
int delay;
if (stream == SNDRV_PCM_STREAM_PLAYBACK)
delay = pos - lpib_pos;
else
delay = lpib_pos - pos;
if (delay < 0) {
if (delay >= azx_dev->core.delay_negative_threshold)
delay = 0;
else
delay += azx_dev->core.bufsize;
}
if (delay >= azx_dev->core.period_bytes) {
dev_info(chip->card->dev,
"Unstable LPIB (%d >= %d); disabling LPIB delay counting\n",
delay, azx_dev->core.period_bytes);
delay = 0;
chip->driver_caps &= ~AZX_DCAPS_COUNT_LPIB_DELAY;
chip->get_delay[stream] = NULL;
}
return bytes_to_frames(substream->runtime, delay);
}
static int azx_position_ok(struct azx *chip, struct azx_dev *azx_dev);
/* called from IRQ */
static int azx_position_check(struct azx *chip, struct azx_dev *azx_dev)
{
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
int ok;
ok = azx_position_ok(chip, azx_dev);
if (ok == 1) {
azx_dev->irq_pending = 0;
return ok;
} else if (ok == 0) {
/* bogus IRQ, process it later */
azx_dev->irq_pending = 1;
schedule_work(&hda->irq_pending_work);
}
return 0;
}
/* Enable/disable i915 display power for the link */
static int azx_intel_link_power(struct azx *chip, bool enable)
{
struct hdac_bus *bus = azx_bus(chip);
return snd_hdac_display_power(bus, enable);
}
/*
* Check whether the current DMA position is acceptable for updating
* periods. Returns non-zero if it's OK.
*
* Many HD-audio controllers appear pretty inaccurate about
* the update-IRQ timing. The IRQ is issued before actually the
* data is processed. So, we need to process it afterwords in a
* workqueue.
*/
static int azx_position_ok(struct azx *chip, struct azx_dev *azx_dev)
{
struct snd_pcm_substream *substream = azx_dev->core.substream;
int stream = substream->stream;
u32 wallclk;
unsigned int pos;
wallclk = azx_readl(chip, WALLCLK) - azx_dev->core.start_wallclk;
if (wallclk < (azx_dev->core.period_wallclk * 2) / 3)
return -1; /* bogus (too early) interrupt */
if (chip->get_position[stream])
pos = chip->get_position[stream](chip, azx_dev);
else { /* use the position buffer as default */
pos = azx_get_pos_posbuf(chip, azx_dev);
if (!pos || pos == (u32)-1) {
dev_info(chip->card->dev,
"Invalid position buffer, using LPIB read method instead.\n");
chip->get_position[stream] = azx_get_pos_lpib;
if (chip->get_position[0] == azx_get_pos_lpib &&
chip->get_position[1] == azx_get_pos_lpib)
azx_bus(chip)->use_posbuf = false;
pos = azx_get_pos_lpib(chip, azx_dev);
chip->get_delay[stream] = NULL;
} else {
chip->get_position[stream] = azx_get_pos_posbuf;
if (chip->driver_caps & AZX_DCAPS_COUNT_LPIB_DELAY)
chip->get_delay[stream] = azx_get_delay_from_lpib;
}
}
if (pos >= azx_dev->core.bufsize)
pos = 0;
if (WARN_ONCE(!azx_dev->core.period_bytes,
"hda-intel: zero azx_dev->period_bytes"))
return -1; /* this shouldn't happen! */
if (wallclk < (azx_dev->core.period_wallclk * 5) / 4 &&
pos % azx_dev->core.period_bytes > azx_dev->core.period_bytes / 2)
/* NG - it's below the first next period boundary */
return chip->bdl_pos_adj ? 0 : -1;
azx_dev->core.start_wallclk += wallclk;
return 1; /* OK, it's fine */
}
/*
* The work for pending PCM period updates.
*/
static void azx_irq_pending_work(struct work_struct *work)
{
struct hda_intel *hda = container_of(work, struct hda_intel, irq_pending_work);
struct azx *chip = &hda->chip;
struct hdac_bus *bus = azx_bus(chip);
struct hdac_stream *s;
int pending, ok;
if (!hda->irq_pending_warned) {
dev_info(chip->card->dev,
"IRQ timing workaround is activated for card #%d. Suggest a bigger bdl_pos_adj.\n",
chip->card->number);
hda->irq_pending_warned = 1;
}
for (;;) {
pending = 0;
spin_lock_irq(&bus->reg_lock);
list_for_each_entry(s, &bus->stream_list, list) {
struct azx_dev *azx_dev = stream_to_azx_dev(s);
if (!azx_dev->irq_pending ||
!s->substream ||
!s->running)
continue;
ok = azx_position_ok(chip, azx_dev);
if (ok > 0) {
azx_dev->irq_pending = 0;
spin_unlock(&bus->reg_lock);
snd_pcm_period_elapsed(s->substream);
spin_lock(&bus->reg_lock);
} else if (ok < 0) {
pending = 0; /* too early */
} else
pending++;
}
spin_unlock_irq(&bus->reg_lock);
if (!pending)
return;
msleep(1);
}
}
/* clear irq_pending flags and assure no on-going workq */
static void azx_clear_irq_pending(struct azx *chip)
{
struct hdac_bus *bus = azx_bus(chip);
struct hdac_stream *s;
spin_lock_irq(&bus->reg_lock);
list_for_each_entry(s, &bus->stream_list, list) {
struct azx_dev *azx_dev = stream_to_azx_dev(s);
azx_dev->irq_pending = 0;
}
spin_unlock_irq(&bus->reg_lock);
}
static int azx_acquire_irq(struct azx *chip, int do_disconnect)
{
struct hdac_bus *bus = azx_bus(chip);
if (request_irq(chip->pci->irq, azx_interrupt,
chip->msi ? 0 : IRQF_SHARED,
chip->card->irq_descr, chip)) {
dev_err(chip->card->dev,
"unable to grab IRQ %d, disabling device\n",
chip->pci->irq);
if (do_disconnect)
snd_card_disconnect(chip->card);
return -1;
}
bus->irq = chip->pci->irq;
pci_intx(chip->pci, !chip->msi);
return 0;
}
/* get the current DMA position with correction on VIA chips */
static unsigned int azx_via_get_position(struct azx *chip,
struct azx_dev *azx_dev)
{
unsigned int link_pos, mini_pos, bound_pos;
unsigned int mod_link_pos, mod_dma_pos, mod_mini_pos;
unsigned int fifo_size;
link_pos = snd_hdac_stream_get_pos_lpib(azx_stream(azx_dev));
if (azx_dev->core.substream->stream == SNDRV_PCM_STREAM_PLAYBACK) {
/* Playback, no problem using link position */
return link_pos;
}
/* Capture */
/* For new chipset,
* use mod to get the DMA position just like old chipset
*/
mod_dma_pos = le32_to_cpu(*azx_dev->core.posbuf);
mod_dma_pos %= azx_dev->core.period_bytes;
/* azx_dev->fifo_size can't get FIFO size of in stream.
* Get from base address + offset.
*/
fifo_size = readw(azx_bus(chip)->remap_addr +
VIA_IN_STREAM0_FIFO_SIZE_OFFSET);
if (azx_dev->insufficient) {
/* Link position never gather than FIFO size */
if (link_pos <= fifo_size)
return 0;
azx_dev->insufficient = 0;
}
if (link_pos <= fifo_size)
mini_pos = azx_dev->core.bufsize + link_pos - fifo_size;
else
mini_pos = link_pos - fifo_size;
/* Find nearest previous boudary */
mod_mini_pos = mini_pos % azx_dev->core.period_bytes;
mod_link_pos = link_pos % azx_dev->core.period_bytes;
if (mod_link_pos >= fifo_size)
bound_pos = link_pos - mod_link_pos;
else if (mod_dma_pos >= mod_mini_pos)
bound_pos = mini_pos - mod_mini_pos;
else {
bound_pos = mini_pos - mod_mini_pos + azx_dev->core.period_bytes;
if (bound_pos >= azx_dev->core.bufsize)
bound_pos = 0;
}
/* Calculate real DMA position we want */
return bound_pos + mod_dma_pos;
}
static unsigned int azx_skl_get_dpib_pos(struct azx *chip,
struct azx_dev *azx_dev)
{
return _snd_hdac_chip_readl(azx_bus(chip),
AZX_REG_VS_SDXDPIB_XBASE +
(AZX_REG_VS_SDXDPIB_XINTERVAL *
azx_dev->core.index));
}
/* get the current DMA position with correction on SKL+ chips */
static unsigned int azx_get_pos_skl(struct azx *chip, struct azx_dev *azx_dev)
{
/* DPIB register gives a more accurate position for playback */
if (azx_dev->core.substream->stream == SNDRV_PCM_STREAM_PLAYBACK)
return azx_skl_get_dpib_pos(chip, azx_dev);
/* For capture, we need to read posbuf, but it requires a delay
* for the possible boundary overlap; the read of DPIB fetches the
* actual posbuf
*/
udelay(20);
azx_skl_get_dpib_pos(chip, azx_dev);
return azx_get_pos_posbuf(chip, azx_dev);
}
#ifdef CONFIG_PM
static DEFINE_MUTEX(card_list_lock);
static LIST_HEAD(card_list);
static void azx_add_card_list(struct azx *chip)
{
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
mutex_lock(&card_list_lock);
list_add(&hda->list, &card_list);
mutex_unlock(&card_list_lock);
}
static void azx_del_card_list(struct azx *chip)
{
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
mutex_lock(&card_list_lock);
list_del_init(&hda->list);
mutex_unlock(&card_list_lock);
}
/* trigger power-save check at writing parameter */
static int param_set_xint(const char *val, const struct kernel_param *kp)
{
struct hda_intel *hda;
struct azx *chip;
int prev = power_save;
int ret = param_set_int(val, kp);
if (ret || prev == power_save)
return ret;
mutex_lock(&card_list_lock);
list_for_each_entry(hda, &card_list, list) {
chip = &hda->chip;
if (!hda->probe_continued || chip->disabled)
continue;
snd_hda_set_power_save(&chip->bus, power_save * 1000);
}
mutex_unlock(&card_list_lock);
return 0;
}
#else
#define azx_add_card_list(chip) /* NOP */
#define azx_del_card_list(chip) /* NOP */
#endif /* CONFIG_PM */
#ifdef CONFIG_PM_SLEEP
/*
* power management
*/
static int azx_suspend(struct device *dev)
{
struct snd_card *card = dev_get_drvdata(dev);
struct azx *chip;
struct hda_intel *hda;
struct hdac_bus *bus;
if (!card)
return 0;
chip = card->private_data;
hda = container_of(chip, struct hda_intel, chip);
if (chip->disabled || hda->init_failed || !chip->running)
return 0;
bus = azx_bus(chip);
snd_power_change_state(card, SNDRV_CTL_POWER_D3hot);
azx_clear_irq_pending(chip);
azx_stop_chip(chip);
azx_enter_link_reset(chip);
if (bus->irq >= 0) {
free_irq(bus->irq, chip);
bus->irq = -1;
}
if (chip->msi)
pci_disable_msi(chip->pci);
if ((chip->driver_caps & AZX_DCAPS_I915_POWERWELL)
&& hda->need_i915_power)
snd_hdac_display_power(bus, false);
trace_azx_suspend(chip);
return 0;
}
static int azx_resume(struct device *dev)
{
struct pci_dev *pci = to_pci_dev(dev);
struct snd_card *card = dev_get_drvdata(dev);
struct azx *chip;
struct hda_intel *hda;
struct hdac_bus *bus;
if (!card)
return 0;
chip = card->private_data;
hda = container_of(chip, struct hda_intel, chip);
bus = azx_bus(chip);
if (chip->disabled || hda->init_failed || !chip->running)
return 0;
if (chip->driver_caps & AZX_DCAPS_I915_POWERWELL) {
snd_hdac_display_power(bus, true);
if (hda->need_i915_power)
snd_hdac_i915_set_bclk(bus);
}
if (chip->msi)
if (pci_enable_msi(pci) < 0)
chip->msi = 0;
if (azx_acquire_irq(chip, 1) < 0)
return -EIO;
azx_init_pci(chip);
hda_intel_init_chip(chip, true);
/* power down again for link-controlled chips */
if ((chip->driver_caps & AZX_DCAPS_I915_POWERWELL) &&
!hda->need_i915_power)
snd_hdac_display_power(bus, false);
snd_power_change_state(card, SNDRV_CTL_POWER_D0);
trace_azx_resume(chip);
return 0;
}
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
/* put codec down to D3 at hibernation for Intel SKL+;
* otherwise BIOS may still access the codec and screw up the driver
*/
static int azx_freeze_noirq(struct device *dev)
{
struct snd_card *card = dev_get_drvdata(dev);
struct azx *chip = card->private_data;
struct pci_dev *pci = to_pci_dev(dev);
if (chip->driver_type == AZX_DRIVER_SKL)
pci_set_power_state(pci, PCI_D3hot);
return 0;
}
static int azx_thaw_noirq(struct device *dev)
{
struct snd_card *card = dev_get_drvdata(dev);
struct azx *chip = card->private_data;
struct pci_dev *pci = to_pci_dev(dev);
if (chip->driver_type == AZX_DRIVER_SKL)
pci_set_power_state(pci, PCI_D0);
return 0;
}
#endif /* CONFIG_PM_SLEEP */
#ifdef CONFIG_PM
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
static int azx_runtime_suspend(struct device *dev)
{
struct snd_card *card = dev_get_drvdata(dev);
struct azx *chip;
struct hda_intel *hda;
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
if (!card)
return 0;
chip = card->private_data;
hda = container_of(chip, struct hda_intel, chip);
if (chip->disabled || hda->init_failed)
snd/hda: add runtime suspend/resume on optimus support (v4) Add support for HDMI audio device on VGA cards that powerdown to D3cold using non-standard ACPI/PCI infrastructure (optimus). This does a couple of things to make it work: a) add a set of power ops for the hdmi domain, and enables them via vga_switcheroo when we are a switcheroo controlled card. This just replaces the runtime resume operation so that when the card is in D3cold the userspace pci config space access via sysfs, the vga switcheroon runtime resume gets called first and it calls the GPU resume callback before calling the sound card runtime resume. b) standard ACPI/PCI stacks won't put a device into D3cold without an ACPI handle, but since the hdmi audio devices on gpus don't have an ACPI handle, we need to manually force the device into D3cold after suspend from the switcheroo path only. c) don't try and do runtime s/r when the GPU is off. d) call runtime suspend/resume during switcheroo suspend/resume this is to make sure the runtime stack knows to try and resume the hdmi audio device for pci config space access. v2: fix incorrect runtime call suspend->resume. v3: rework irq handler to avoid false irq when we are resuming but haven't runtime resumed yet, don't bother trying D3cold, it won't work, just set it manually ourselves, move runtime s/r calls outside the main s/r hook. enable dnyamic pm properly by dropping reference. v4: put back irq handler check just wrap it with cap check Acked-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-07-29 05:19:29 +00:00
return 0;
if (!azx_has_pm_runtime(chip))
snd/hda: add runtime suspend/resume on optimus support (v4) Add support for HDMI audio device on VGA cards that powerdown to D3cold using non-standard ACPI/PCI infrastructure (optimus). This does a couple of things to make it work: a) add a set of power ops for the hdmi domain, and enables them via vga_switcheroo when we are a switcheroo controlled card. This just replaces the runtime resume operation so that when the card is in D3cold the userspace pci config space access via sysfs, the vga switcheroon runtime resume gets called first and it calls the GPU resume callback before calling the sound card runtime resume. b) standard ACPI/PCI stacks won't put a device into D3cold without an ACPI handle, but since the hdmi audio devices on gpus don't have an ACPI handle, we need to manually force the device into D3cold after suspend from the switcheroo path only. c) don't try and do runtime s/r when the GPU is off. d) call runtime suspend/resume during switcheroo suspend/resume this is to make sure the runtime stack knows to try and resume the hdmi audio device for pci config space access. v2: fix incorrect runtime call suspend->resume. v3: rework irq handler to avoid false irq when we are resuming but haven't runtime resumed yet, don't bother trying D3cold, it won't work, just set it manually ourselves, move runtime s/r calls outside the main s/r hook. enable dnyamic pm properly by dropping reference. v4: put back irq handler check just wrap it with cap check Acked-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-07-29 05:19:29 +00:00
return 0;
/* enable controller wake up event */
azx_writew(chip, WAKEEN, azx_readw(chip, WAKEEN) |
STATESTS_INT_MASK);
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
azx_stop_chip(chip);
azx_enter_link_reset(chip);
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
azx_clear_irq_pending(chip);
if ((chip->driver_caps & AZX_DCAPS_I915_POWERWELL)
&& hda->need_i915_power)
snd_hdac_display_power(azx_bus(chip), false);
trace_azx_runtime_suspend(chip);
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
return 0;
}
static int azx_runtime_resume(struct device *dev)
{
struct snd_card *card = dev_get_drvdata(dev);
struct azx *chip;
struct hda_intel *hda;
struct hdac_bus *bus;
struct hda_codec *codec;
int status;
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
if (!card)
return 0;
chip = card->private_data;
hda = container_of(chip, struct hda_intel, chip);
bus = azx_bus(chip);
if (chip->disabled || hda->init_failed)
snd/hda: add runtime suspend/resume on optimus support (v4) Add support for HDMI audio device on VGA cards that powerdown to D3cold using non-standard ACPI/PCI infrastructure (optimus). This does a couple of things to make it work: a) add a set of power ops for the hdmi domain, and enables them via vga_switcheroo when we are a switcheroo controlled card. This just replaces the runtime resume operation so that when the card is in D3cold the userspace pci config space access via sysfs, the vga switcheroon runtime resume gets called first and it calls the GPU resume callback before calling the sound card runtime resume. b) standard ACPI/PCI stacks won't put a device into D3cold without an ACPI handle, but since the hdmi audio devices on gpus don't have an ACPI handle, we need to manually force the device into D3cold after suspend from the switcheroo path only. c) don't try and do runtime s/r when the GPU is off. d) call runtime suspend/resume during switcheroo suspend/resume this is to make sure the runtime stack knows to try and resume the hdmi audio device for pci config space access. v2: fix incorrect runtime call suspend->resume. v3: rework irq handler to avoid false irq when we are resuming but haven't runtime resumed yet, don't bother trying D3cold, it won't work, just set it manually ourselves, move runtime s/r calls outside the main s/r hook. enable dnyamic pm properly by dropping reference. v4: put back irq handler check just wrap it with cap check Acked-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-07-29 05:19:29 +00:00
return 0;
if (!azx_has_pm_runtime(chip))
snd/hda: add runtime suspend/resume on optimus support (v4) Add support for HDMI audio device on VGA cards that powerdown to D3cold using non-standard ACPI/PCI infrastructure (optimus). This does a couple of things to make it work: a) add a set of power ops for the hdmi domain, and enables them via vga_switcheroo when we are a switcheroo controlled card. This just replaces the runtime resume operation so that when the card is in D3cold the userspace pci config space access via sysfs, the vga switcheroon runtime resume gets called first and it calls the GPU resume callback before calling the sound card runtime resume. b) standard ACPI/PCI stacks won't put a device into D3cold without an ACPI handle, but since the hdmi audio devices on gpus don't have an ACPI handle, we need to manually force the device into D3cold after suspend from the switcheroo path only. c) don't try and do runtime s/r when the GPU is off. d) call runtime suspend/resume during switcheroo suspend/resume this is to make sure the runtime stack knows to try and resume the hdmi audio device for pci config space access. v2: fix incorrect runtime call suspend->resume. v3: rework irq handler to avoid false irq when we are resuming but haven't runtime resumed yet, don't bother trying D3cold, it won't work, just set it manually ourselves, move runtime s/r calls outside the main s/r hook. enable dnyamic pm properly by dropping reference. v4: put back irq handler check just wrap it with cap check Acked-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-07-29 05:19:29 +00:00
return 0;
if (chip->driver_caps & AZX_DCAPS_I915_POWERWELL) {
snd_hdac_display_power(bus, true);
if (hda->need_i915_power)
snd_hdac_i915_set_bclk(bus);
}
/* Read STATESTS before controller reset */
status = azx_readw(chip, STATESTS);
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
azx_init_pci(chip);
hda_intel_init_chip(chip, true);
if (status) {
list_for_each_codec(codec, &chip->bus)
if (status & (1 << codec->addr))
schedule_delayed_work(&codec->jackpoll_work,
codec->jackpoll_interval);
}
/* disable controller Wake Up event*/
azx_writew(chip, WAKEEN, azx_readw(chip, WAKEEN) &
~STATESTS_INT_MASK);
/* power down again for link-controlled chips */
if ((chip->driver_caps & AZX_DCAPS_I915_POWERWELL) &&
!hda->need_i915_power)
snd_hdac_display_power(bus, false);
trace_azx_runtime_resume(chip);
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
return 0;
}
static int azx_runtime_idle(struct device *dev)
{
struct snd_card *card = dev_get_drvdata(dev);
struct azx *chip;
struct hda_intel *hda;
if (!card)
return 0;
chip = card->private_data;
hda = container_of(chip, struct hda_intel, chip);
if (chip->disabled || hda->init_failed)
snd/hda: add runtime suspend/resume on optimus support (v4) Add support for HDMI audio device on VGA cards that powerdown to D3cold using non-standard ACPI/PCI infrastructure (optimus). This does a couple of things to make it work: a) add a set of power ops for the hdmi domain, and enables them via vga_switcheroo when we are a switcheroo controlled card. This just replaces the runtime resume operation so that when the card is in D3cold the userspace pci config space access via sysfs, the vga switcheroon runtime resume gets called first and it calls the GPU resume callback before calling the sound card runtime resume. b) standard ACPI/PCI stacks won't put a device into D3cold without an ACPI handle, but since the hdmi audio devices on gpus don't have an ACPI handle, we need to manually force the device into D3cold after suspend from the switcheroo path only. c) don't try and do runtime s/r when the GPU is off. d) call runtime suspend/resume during switcheroo suspend/resume this is to make sure the runtime stack knows to try and resume the hdmi audio device for pci config space access. v2: fix incorrect runtime call suspend->resume. v3: rework irq handler to avoid false irq when we are resuming but haven't runtime resumed yet, don't bother trying D3cold, it won't work, just set it manually ourselves, move runtime s/r calls outside the main s/r hook. enable dnyamic pm properly by dropping reference. v4: put back irq handler check just wrap it with cap check Acked-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-07-29 05:19:29 +00:00
return 0;
if (!power_save_controller || !azx_has_pm_runtime(chip) ||
azx_bus(chip)->codec_powered || !chip->running)
return -EBUSY;
return 0;
}
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
static const struct dev_pm_ops azx_pm = {
SET_SYSTEM_SLEEP_PM_OPS(azx_suspend, azx_resume)
#ifdef CONFIG_PM_SLEEP
.freeze_noirq = azx_freeze_noirq,
.thaw_noirq = azx_thaw_noirq,
#endif
SET_RUNTIME_PM_OPS(azx_runtime_suspend, azx_runtime_resume, azx_runtime_idle)
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
};
#define AZX_PM_OPS &azx_pm
#else
#define AZX_PM_OPS NULL
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
#endif /* CONFIG_PM */
static int azx_probe_continue(struct azx *chip);
#ifdef SUPPORT_VGA_SWITCHEROO
static struct pci_dev *get_bound_vga(struct pci_dev *pci);
static void azx_vs_set_state(struct pci_dev *pci,
enum vga_switcheroo_state state)
{
struct snd_card *card = pci_get_drvdata(pci);
struct azx *chip = card->private_data;
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
struct hda_codec *codec;
bool disabled;
wait_for_completion(&hda->probe_wait);
if (hda->init_failed)
return;
disabled = (state == VGA_SWITCHEROO_OFF);
if (chip->disabled == disabled)
return;
if (!hda->probe_continued) {
chip->disabled = disabled;
if (!disabled) {
dev_info(chip->card->dev,
"Start delayed initialization\n");
if (azx_probe_continue(chip) < 0) {
dev_err(chip->card->dev, "initialization error\n");
hda->init_failed = true;
}
}
} else {
dev_info(chip->card->dev, "%s via vga_switcheroo\n",
disabled ? "Disabling" : "Enabling");
if (disabled) {
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
list_for_each_codec(codec, &chip->bus) {
pm_runtime_suspend(hda_codec_dev(codec));
pm_runtime_disable(hda_codec_dev(codec));
}
pm_runtime_suspend(card->dev);
pm_runtime_disable(card->dev);
/* when we get suspended by vga_switcheroo we end up in D3cold,
snd/hda: add runtime suspend/resume on optimus support (v4) Add support for HDMI audio device on VGA cards that powerdown to D3cold using non-standard ACPI/PCI infrastructure (optimus). This does a couple of things to make it work: a) add a set of power ops for the hdmi domain, and enables them via vga_switcheroo when we are a switcheroo controlled card. This just replaces the runtime resume operation so that when the card is in D3cold the userspace pci config space access via sysfs, the vga switcheroon runtime resume gets called first and it calls the GPU resume callback before calling the sound card runtime resume. b) standard ACPI/PCI stacks won't put a device into D3cold without an ACPI handle, but since the hdmi audio devices on gpus don't have an ACPI handle, we need to manually force the device into D3cold after suspend from the switcheroo path only. c) don't try and do runtime s/r when the GPU is off. d) call runtime suspend/resume during switcheroo suspend/resume this is to make sure the runtime stack knows to try and resume the hdmi audio device for pci config space access. v2: fix incorrect runtime call suspend->resume. v3: rework irq handler to avoid false irq when we are resuming but haven't runtime resumed yet, don't bother trying D3cold, it won't work, just set it manually ourselves, move runtime s/r calls outside the main s/r hook. enable dnyamic pm properly by dropping reference. v4: put back irq handler check just wrap it with cap check Acked-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-07-29 05:19:29 +00:00
* however we have no ACPI handle, so pci/acpi can't put us there,
* put ourselves there */
pci->current_state = PCI_D3cold;
chip->disabled = true;
if (snd_hda_lock_devices(&chip->bus))
dev_warn(chip->card->dev,
"Cannot lock devices!\n");
} else {
snd_hda_unlock_devices(&chip->bus);
chip->disabled = false;
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
pm_runtime_enable(card->dev);
list_for_each_codec(codec, &chip->bus) {
pm_runtime_enable(hda_codec_dev(codec));
pm_runtime_resume(hda_codec_dev(codec));
}
}
}
}
static bool azx_vs_can_switch(struct pci_dev *pci)
{
struct snd_card *card = pci_get_drvdata(pci);
struct azx *chip = card->private_data;
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
wait_for_completion(&hda->probe_wait);
if (hda->init_failed)
return false;
if (chip->disabled || !hda->probe_continued)
return true;
if (snd_hda_lock_devices(&chip->bus))
return false;
snd_hda_unlock_devices(&chip->bus);
return true;
}
static void init_vga_switcheroo(struct azx *chip)
{
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
struct pci_dev *p = get_bound_vga(chip->pci);
if (p) {
dev_info(chip->card->dev,
"Handle vga_switcheroo audio client\n");
hda->use_vga_switcheroo = 1;
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
chip->driver_caps |= AZX_DCAPS_PM_RUNTIME;
pci_dev_put(p);
}
}
static const struct vga_switcheroo_client_ops azx_vs_ops = {
.set_gpu_state = azx_vs_set_state,
.can_switch = azx_vs_can_switch,
};
static int register_vga_switcheroo(struct azx *chip)
{
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
struct pci_dev *p;
int err;
if (!hda->use_vga_switcheroo)
return 0;
p = get_bound_vga(chip->pci);
err = vga_switcheroo_register_audio_client(chip->pci, &azx_vs_ops, p);
pci_dev_put(p);
if (err < 0)
return err;
hda->vga_switcheroo_registered = 1;
snd/hda: add runtime suspend/resume on optimus support (v4) Add support for HDMI audio device on VGA cards that powerdown to D3cold using non-standard ACPI/PCI infrastructure (optimus). This does a couple of things to make it work: a) add a set of power ops for the hdmi domain, and enables them via vga_switcheroo when we are a switcheroo controlled card. This just replaces the runtime resume operation so that when the card is in D3cold the userspace pci config space access via sysfs, the vga switcheroon runtime resume gets called first and it calls the GPU resume callback before calling the sound card runtime resume. b) standard ACPI/PCI stacks won't put a device into D3cold without an ACPI handle, but since the hdmi audio devices on gpus don't have an ACPI handle, we need to manually force the device into D3cold after suspend from the switcheroo path only. c) don't try and do runtime s/r when the GPU is off. d) call runtime suspend/resume during switcheroo suspend/resume this is to make sure the runtime stack knows to try and resume the hdmi audio device for pci config space access. v2: fix incorrect runtime call suspend->resume. v3: rework irq handler to avoid false irq when we are resuming but haven't runtime resumed yet, don't bother trying D3cold, it won't work, just set it manually ourselves, move runtime s/r calls outside the main s/r hook. enable dnyamic pm properly by dropping reference. v4: put back irq handler check just wrap it with cap check Acked-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-07-29 05:19:29 +00:00
return 0;
}
#else
#define init_vga_switcheroo(chip) /* NOP */
#define register_vga_switcheroo(chip) 0
#define check_hdmi_disabled(pci) false
#endif /* SUPPORT_VGA_SWITCHER */
/*
* destructor
*/
static int azx_free(struct azx *chip)
{
struct pci_dev *pci = chip->pci;
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
struct hdac_bus *bus = azx_bus(chip);
if (azx_has_pm_runtime(chip) && chip->running)
pm_runtime_get_noresume(&pci->dev);
azx_del_card_list(chip);
hda->init_failed = 1; /* to be sure */
complete_all(&hda->probe_wait);
if (use_vga_switcheroo(hda)) {
if (chip->disabled && hda->probe_continued)
snd_hda_unlock_devices(&chip->bus);
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
if (hda->vga_switcheroo_registered)
vga_switcheroo_unregister_client(chip->pci);
}
if (bus->chip_init) {
azx_clear_irq_pending(chip);
azx_stop_all_streams(chip);
azx_stop_chip(chip);
}
if (bus->irq >= 0)
free_irq(bus->irq, (void*)chip);
if (chip->msi)
pci_disable_msi(chip->pci);
iounmap(bus->remap_addr);
azx_free_stream_pages(chip);
azx_free_streams(chip);
snd_hdac_bus_exit(bus);
if (chip->region_requested)
pci_release_regions(chip->pci);
pci_disable_device(chip->pci);
#ifdef CONFIG_SND_HDA_PATCH_LOADER
release_firmware(chip->fw);
#endif
if (chip->driver_caps & AZX_DCAPS_I915_POWERWELL) {
if (hda->need_i915_power)
snd_hdac_display_power(bus, false);
}
if (chip->driver_caps & AZX_DCAPS_I915_COMPONENT)
snd_hdac_i915_exit(bus);
kfree(hda);
return 0;
}
static int azx_dev_disconnect(struct snd_device *device)
{
struct azx *chip = device->device_data;
chip->bus.shutdown = 1;
return 0;
}
static int azx_dev_free(struct snd_device *device)
{
return azx_free(device->device_data);
}
#ifdef SUPPORT_VGA_SWITCHEROO
/*
* Check of disabled HDMI controller by vga_switcheroo
*/
static struct pci_dev *get_bound_vga(struct pci_dev *pci)
{
struct pci_dev *p;
/* check only discrete GPU */
switch (pci->vendor) {
case PCI_VENDOR_ID_ATI:
case PCI_VENDOR_ID_AMD:
case PCI_VENDOR_ID_NVIDIA:
if (pci->devfn == 1) {
p = pci_get_domain_bus_and_slot(pci_domain_nr(pci->bus),
pci->bus->number, 0);
if (p) {
if ((p->class >> 16) == PCI_BASE_CLASS_DISPLAY)
return p;
pci_dev_put(p);
}
}
break;
}
return NULL;
}
static bool check_hdmi_disabled(struct pci_dev *pci)
{
bool vga_inactive = false;
struct pci_dev *p = get_bound_vga(pci);
if (p) {
if (vga_switcheroo_get_client_state(p) == VGA_SWITCHEROO_OFF)
vga_inactive = true;
pci_dev_put(p);
}
return vga_inactive;
}
#endif /* SUPPORT_VGA_SWITCHEROO */
/*
* white/black-listing for position_fix
*/
static struct snd_pci_quirk position_fix_list[] = {
SND_PCI_QUIRK(0x1028, 0x01cc, "Dell D820", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1028, 0x01de, "Dell Precision 390", POS_FIX_LPIB),
SND_PCI_QUIRK(0x103c, 0x306d, "HP dv3", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1043, 0x813d, "ASUS P5AD2", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1043, 0x81b3, "ASUS", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1043, 0x81e7, "ASUS M2V", POS_FIX_LPIB),
SND_PCI_QUIRK(0x104d, 0x9069, "Sony VPCS11V9E", POS_FIX_LPIB),
SND_PCI_QUIRK(0x10de, 0xcb89, "Macbook Pro 7,1", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1297, 0x3166, "Shuttle", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1458, 0xa022, "ga-ma770-ud3", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1462, 0x1002, "MSI Wind U115", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1565, 0x8218, "Biostar Microtech", POS_FIX_LPIB),
SND_PCI_QUIRK(0x1849, 0x0888, "775Dual-VSTA", POS_FIX_LPIB),
SND_PCI_QUIRK(0x8086, 0x2503, "DG965OT AAD63733-203", POS_FIX_LPIB),
{}
};
static int check_position_fix(struct azx *chip, int fix)
{
const struct snd_pci_quirk *q;
switch (fix) {
case POS_FIX_AUTO:
case POS_FIX_LPIB:
case POS_FIX_POSBUF:
case POS_FIX_VIACOMBO:
case POS_FIX_COMBO:
case POS_FIX_SKL:
return fix;
}
q = snd_pci_quirk_lookup(chip->pci, position_fix_list);
if (q) {
dev_info(chip->card->dev,
"position_fix set to %d for device %04x:%04x\n",
q->value, q->subvendor, q->subdevice);
return q->value;
}
/* Check VIA/ATI HD Audio Controller exist */
if (chip->driver_type == AZX_DRIVER_VIA) {
dev_dbg(chip->card->dev, "Using VIACOMBO position fix\n");
return POS_FIX_VIACOMBO;
}
if (chip->driver_caps & AZX_DCAPS_POSFIX_LPIB) {
dev_dbg(chip->card->dev, "Using LPIB position fix\n");
return POS_FIX_LPIB;
}
if (chip->driver_type == AZX_DRIVER_SKL) {
dev_dbg(chip->card->dev, "Using SKL position fix\n");
return POS_FIX_SKL;
}
return POS_FIX_AUTO;
}
static void assign_position_fix(struct azx *chip, int fix)
{
static azx_get_pos_callback_t callbacks[] = {
[POS_FIX_AUTO] = NULL,
[POS_FIX_LPIB] = azx_get_pos_lpib,
[POS_FIX_POSBUF] = azx_get_pos_posbuf,
[POS_FIX_VIACOMBO] = azx_via_get_position,
[POS_FIX_COMBO] = azx_get_pos_lpib,
[POS_FIX_SKL] = azx_get_pos_skl,
};
chip->get_position[0] = chip->get_position[1] = callbacks[fix];
/* combo mode uses LPIB only for playback */
if (fix == POS_FIX_COMBO)
chip->get_position[1] = NULL;
if ((fix == POS_FIX_POSBUF || fix == POS_FIX_SKL) &&
(chip->driver_caps & AZX_DCAPS_COUNT_LPIB_DELAY)) {
chip->get_delay[0] = chip->get_delay[1] =
azx_get_delay_from_lpib;
}
}
/*
* black-lists for probe_mask
*/
static struct snd_pci_quirk probe_mask_list[] = {
/* Thinkpad often breaks the controller communication when accessing
* to the non-working (or non-existing) modem codec slot.
*/
SND_PCI_QUIRK(0x1014, 0x05b7, "Thinkpad Z60", 0x01),
SND_PCI_QUIRK(0x17aa, 0x2010, "Thinkpad X/T/R60", 0x01),
SND_PCI_QUIRK(0x17aa, 0x20ac, "Thinkpad X/T/R61", 0x01),
/* broken BIOS */
SND_PCI_QUIRK(0x1028, 0x20ac, "Dell Studio Desktop", 0x01),
/* including bogus ALC268 in slot#2 that conflicts with ALC888 */
SND_PCI_QUIRK(0x17c0, 0x4085, "Medion MD96630", 0x01),
/* forced codec slots */
SND_PCI_QUIRK(0x1043, 0x1262, "ASUS W5Fm", 0x103),
SND_PCI_QUIRK(0x1046, 0x1262, "ASUS W5F", 0x103),
/* WinFast VP200 H (Teradici) user reported broken communication */
SND_PCI_QUIRK(0x3a21, 0x040d, "WinFast VP200 H", 0x101),
{}
};
#define AZX_FORCE_CODEC_MASK 0x100
static void check_probe_mask(struct azx *chip, int dev)
{
const struct snd_pci_quirk *q;
chip->codec_probe_mask = probe_mask[dev];
if (chip->codec_probe_mask == -1) {
q = snd_pci_quirk_lookup(chip->pci, probe_mask_list);
if (q) {
dev_info(chip->card->dev,
"probe_mask set to 0x%x for device %04x:%04x\n",
q->value, q->subvendor, q->subdevice);
chip->codec_probe_mask = q->value;
}
}
/* check forced option */
if (chip->codec_probe_mask != -1 &&
(chip->codec_probe_mask & AZX_FORCE_CODEC_MASK)) {
azx_bus(chip)->codec_mask = chip->codec_probe_mask & 0xff;
dev_info(chip->card->dev, "codec_mask forced to 0x%x\n",
(int)azx_bus(chip)->codec_mask);
}
}
/*
* white/black-list for enable_msi
*/
static struct snd_pci_quirk msi_black_list[] = {
SND_PCI_QUIRK(0x103c, 0x2191, "HP", 0), /* AMD Hudson */
SND_PCI_QUIRK(0x103c, 0x2192, "HP", 0), /* AMD Hudson */
SND_PCI_QUIRK(0x103c, 0x21f7, "HP", 0), /* AMD Hudson */
SND_PCI_QUIRK(0x103c, 0x21fa, "HP", 0), /* AMD Hudson */
SND_PCI_QUIRK(0x1043, 0x81f2, "ASUS", 0), /* Athlon64 X2 + nvidia */
SND_PCI_QUIRK(0x1043, 0x81f6, "ASUS", 0), /* nvidia */
SND_PCI_QUIRK(0x1043, 0x822d, "ASUS", 0), /* Athlon64 X2 + nvidia MCP55 */
SND_PCI_QUIRK(0x1179, 0xfb44, "Toshiba Satellite C870", 0), /* AMD Hudson */
SND_PCI_QUIRK(0x1849, 0x0888, "ASRock", 0), /* Athlon64 X2 + nvidia */
SND_PCI_QUIRK(0xa0a0, 0x0575, "Aopen MZ915-M", 0), /* ICH6 */
{}
};
static void check_msi(struct azx *chip)
{
const struct snd_pci_quirk *q;
if (enable_msi >= 0) {
chip->msi = !!enable_msi;
return;
}
chip->msi = 1; /* enable MSI as default */
q = snd_pci_quirk_lookup(chip->pci, msi_black_list);
if (q) {
dev_info(chip->card->dev,
"msi for device %04x:%04x set to %d\n",
q->subvendor, q->subdevice, q->value);
chip->msi = q->value;
return;
}
/* NVidia chipsets seem to cause troubles with MSI */
if (chip->driver_caps & AZX_DCAPS_NO_MSI) {
dev_info(chip->card->dev, "Disabling MSI\n");
chip->msi = 0;
}
}
/* check the snoop mode availability */
static void azx_check_snoop_available(struct azx *chip)
{
int snoop = hda_snoop;
if (snoop >= 0) {
dev_info(chip->card->dev, "Force to %s mode by module option\n",
snoop ? "snoop" : "non-snoop");
chip->snoop = snoop;
return;
}
snoop = true;
if (azx_get_snoop_type(chip) == AZX_SNOOP_TYPE_NONE &&
chip->driver_type == AZX_DRIVER_VIA) {
/* force to non-snoop mode for a new VIA controller
* when BIOS is set
*/
u8 val;
pci_read_config_byte(chip->pci, 0x42, &val);
if (!(val & 0x80) && (chip->pci->revision == 0x30 ||
chip->pci->revision == 0x20))
snoop = false;
}
if (chip->driver_caps & AZX_DCAPS_SNOOP_OFF)
snoop = false;
chip->snoop = snoop;
if (!snoop)
dev_info(chip->card->dev, "Force to non-snoop mode\n");
}
static void azx_probe_work(struct work_struct *work)
{
struct hda_intel *hda = container_of(work, struct hda_intel, probe_work);
azx_probe_continue(&hda->chip);
}
static int default_bdl_pos_adj(struct azx *chip)
{
/* some exceptions: Atoms seem problematic with value 1 */
if (chip->pci->vendor == PCI_VENDOR_ID_INTEL) {
switch (chip->pci->device) {
case 0x0f04: /* Baytrail */
case 0x2284: /* Braswell */
return 32;
}
}
switch (chip->driver_type) {
case AZX_DRIVER_ICH:
case AZX_DRIVER_PCH:
return 1;
default:
return 32;
}
}
/*
* constructor
*/
static const struct hdac_io_ops pci_hda_io_ops;
static const struct hda_controller_ops pci_hda_ops;
static int azx_create(struct snd_card *card, struct pci_dev *pci,
int dev, unsigned int driver_caps,
struct azx **rchip)
{
static struct snd_device_ops ops = {
.dev_disconnect = azx_dev_disconnect,
.dev_free = azx_dev_free,
};
struct hda_intel *hda;
struct azx *chip;
int err;
*rchip = NULL;
err = pci_enable_device(pci);
if (err < 0)
return err;
hda = kzalloc(sizeof(*hda), GFP_KERNEL);
if (!hda) {
pci_disable_device(pci);
return -ENOMEM;
}
chip = &hda->chip;
mutex_init(&chip->open_mutex);
chip->card = card;
chip->pci = pci;
chip->ops = &pci_hda_ops;
chip->driver_caps = driver_caps;
chip->driver_type = driver_caps & 0xff;
check_msi(chip);
chip->dev_index = dev;
chip->jackpoll_ms = jackpoll_ms;
INIT_LIST_HEAD(&chip->pcm_list);
INIT_WORK(&hda->irq_pending_work, azx_irq_pending_work);
INIT_LIST_HEAD(&hda->list);
init_vga_switcheroo(chip);
init_completion(&hda->probe_wait);
assign_position_fix(chip, check_position_fix(chip, position_fix[dev]));
check_probe_mask(chip, dev);
if (single_cmd < 0) /* allow fallback to single_cmd at errors */
chip->fallback_to_single_cmd = 1;
else /* explicitly set to single_cmd or not */
chip->single_cmd = single_cmd;
azx_check_snoop_available(chip);
if (bdl_pos_adj[dev] < 0)
chip->bdl_pos_adj = default_bdl_pos_adj(chip);
else
chip->bdl_pos_adj = bdl_pos_adj[dev];
/* Workaround for a communication error on CFL (bko#199007) */
if (IS_CFL(pci))
chip->polling_mode = 1;
err = azx_bus_init(chip, model[dev], &pci_hda_io_ops);
if (err < 0) {
kfree(hda);
pci_disable_device(pci);
return err;
}
if (chip->driver_type == AZX_DRIVER_NVIDIA) {
dev_dbg(chip->card->dev, "Enable delay in RIRB handling\n");
chip->bus.needs_damn_long_delay = 1;
}
err = snd_device_new(card, SNDRV_DEV_LOWLEVEL, chip, &ops);
if (err < 0) {
dev_err(card->dev, "Error creating device [card]!\n");
azx_free(chip);
return err;
}
/* continue probing in work context as may trigger request module */
INIT_WORK(&hda->probe_work, azx_probe_work);
*rchip = chip;
return 0;
}
static int azx_first_init(struct azx *chip)
{
int dev = chip->dev_index;
struct pci_dev *pci = chip->pci;
struct snd_card *card = chip->card;
struct hdac_bus *bus = azx_bus(chip);
int err;
unsigned short gcap;
unsigned int dma_bits = 64;
#if BITS_PER_LONG != 64
/* Fix up base address on ULI M5461 */
if (chip->driver_type == AZX_DRIVER_ULI) {
u16 tmp3;
pci_read_config_word(pci, 0x40, &tmp3);
pci_write_config_word(pci, 0x40, tmp3 | 0x10);
pci_write_config_dword(pci, PCI_BASE_ADDRESS_1, 0);
}
#endif
err = pci_request_regions(pci, "ICH HD audio");
if (err < 0)
return err;
chip->region_requested = 1;
bus->addr = pci_resource_start(pci, 0);
bus->remap_addr = pci_ioremap_bar(pci, 0);
if (bus->remap_addr == NULL) {
dev_err(card->dev, "ioremap error\n");
return -ENXIO;
}
if (chip->driver_type == AZX_DRIVER_SKL)
snd_hdac_bus_parse_capabilities(bus);
/*
* Some Intel CPUs has always running timer (ART) feature and
* controller may have Global time sync reporting capability, so
* check both of these before declaring synchronized time reporting
* capability SNDRV_PCM_INFO_HAS_LINK_SYNCHRONIZED_ATIME
*/
chip->gts_present = false;
#ifdef CONFIG_X86
if (bus->ppcap && boot_cpu_has(X86_FEATURE_ART))
chip->gts_present = true;
#endif
if (chip->msi) {
if (chip->driver_caps & AZX_DCAPS_NO_MSI64) {
dev_dbg(card->dev, "Disabling 64bit MSI\n");
pci->no_64bit_msi = true;
}
if (pci_enable_msi(pci) < 0)
chip->msi = 0;
}
if (azx_acquire_irq(chip, 0) < 0)
return -EBUSY;
pci_set_master(pci);
synchronize_irq(bus->irq);
gcap = azx_readw(chip, GCAP);
dev_dbg(card->dev, "chipset global capabilities = 0x%x\n", gcap);
/* AMD devices support 40 or 48bit DMA, take the safe one */
if (chip->pci->vendor == PCI_VENDOR_ID_AMD)
dma_bits = 40;
/* disable SB600 64bit support for safety */
if (chip->pci->vendor == PCI_VENDOR_ID_ATI) {
struct pci_dev *p_smbus;
dma_bits = 40;
p_smbus = pci_get_device(PCI_VENDOR_ID_ATI,
PCI_DEVICE_ID_ATI_SBX00_SMBUS,
NULL);
if (p_smbus) {
if (p_smbus->revision < 0x30)
gcap &= ~AZX_GCAP_64OK;
pci_dev_put(p_smbus);
}
}
/* NVidia hardware normally only supports up to 40 bits of DMA */
if (chip->pci->vendor == PCI_VENDOR_ID_NVIDIA)
dma_bits = 40;
/* disable 64bit DMA address on some devices */
if (chip->driver_caps & AZX_DCAPS_NO_64BIT) {
dev_dbg(card->dev, "Disabling 64bit DMA\n");
gcap &= ~AZX_GCAP_64OK;
}
/* disable buffer size rounding to 128-byte multiples if supported */
if (align_buffer_size >= 0)
chip->align_buffer_size = !!align_buffer_size;
else {
if (chip->driver_caps & AZX_DCAPS_NO_ALIGN_BUFSIZE)
chip->align_buffer_size = 0;
else
chip->align_buffer_size = 1;
}
/* allow 64bit DMA address if supported by H/W */
if (!(gcap & AZX_GCAP_64OK))
dma_bits = 32;
if (!dma_set_mask(&pci->dev, DMA_BIT_MASK(dma_bits))) {
dma_set_coherent_mask(&pci->dev, DMA_BIT_MASK(dma_bits));
} else {
dma_set_mask(&pci->dev, DMA_BIT_MASK(32));
dma_set_coherent_mask(&pci->dev, DMA_BIT_MASK(32));
}
/* read number of streams from GCAP register instead of using
* hardcoded value
*/
chip->capture_streams = (gcap >> 8) & 0x0f;
chip->playback_streams = (gcap >> 12) & 0x0f;
if (!chip->playback_streams && !chip->capture_streams) {
/* gcap didn't give any info, switching to old method */
switch (chip->driver_type) {
case AZX_DRIVER_ULI:
chip->playback_streams = ULI_NUM_PLAYBACK;
chip->capture_streams = ULI_NUM_CAPTURE;
break;
case AZX_DRIVER_ATIHDMI:
case AZX_DRIVER_ATIHDMI_NS:
chip->playback_streams = ATIHDMI_NUM_PLAYBACK;
chip->capture_streams = ATIHDMI_NUM_CAPTURE;
break;
case AZX_DRIVER_GENERIC:
default:
chip->playback_streams = ICH6_NUM_PLAYBACK;
chip->capture_streams = ICH6_NUM_CAPTURE;
break;
}
}
chip->capture_index_offset = 0;
chip->playback_index_offset = chip->capture_streams;
chip->num_streams = chip->playback_streams + chip->capture_streams;
/* sanity check for the SDxCTL.STRM field overflow */
if (chip->num_streams > 15 &&
(chip->driver_caps & AZX_DCAPS_SEPARATE_STREAM_TAG) == 0) {
dev_warn(chip->card->dev, "number of I/O streams is %d, "
"forcing separate stream tags", chip->num_streams);
chip->driver_caps |= AZX_DCAPS_SEPARATE_STREAM_TAG;
}
/* initialize streams */
err = azx_init_streams(chip);
if (err < 0)
return err;
err = azx_alloc_stream_pages(chip);
if (err < 0)
return err;
/* initialize chip */
azx_init_pci(chip);
if (chip->driver_caps & AZX_DCAPS_I915_POWERWELL)
snd_hdac_i915_set_bclk(bus);
hda_intel_init_chip(chip, (probe_only[dev] & 2) == 0);
/* codec detection */
if (!azx_bus(chip)->codec_mask) {
dev_err(card->dev, "no codecs found!\n");
return -ENODEV;
}
strcpy(card->driver, "HDA-Intel");
strlcpy(card->shortname, driver_short_names[chip->driver_type],
sizeof(card->shortname));
snprintf(card->longname, sizeof(card->longname),
"%s at 0x%lx irq %i",
card->shortname, bus->addr, bus->irq);
return 0;
}
#ifdef CONFIG_SND_HDA_PATCH_LOADER
/* callback from request_firmware_nowait() */
static void azx_firmware_cb(const struct firmware *fw, void *context)
{
struct snd_card *card = context;
struct azx *chip = card->private_data;
struct pci_dev *pci = chip->pci;
if (!fw) {
dev_err(card->dev, "Cannot load firmware, aborting\n");
goto error;
}
chip->fw = fw;
if (!chip->disabled) {
/* continue probing */
if (azx_probe_continue(chip))
goto error;
}
return; /* OK */
error:
snd_card_free(card);
pci_set_drvdata(pci, NULL);
}
#endif
/*
* HDA controller ops.
*/
/* PCI register access. */
static void pci_azx_writel(u32 value, u32 __iomem *addr)
{
writel(value, addr);
}
static u32 pci_azx_readl(u32 __iomem *addr)
{
return readl(addr);
}
static void pci_azx_writew(u16 value, u16 __iomem *addr)
{
writew(value, addr);
}
static u16 pci_azx_readw(u16 __iomem *addr)
{
return readw(addr);
}
static void pci_azx_writeb(u8 value, u8 __iomem *addr)
{
writeb(value, addr);
}
static u8 pci_azx_readb(u8 __iomem *addr)
{
return readb(addr);
}
static int disable_msi_reset_irq(struct azx *chip)
{
struct hdac_bus *bus = azx_bus(chip);
int err;
free_irq(bus->irq, chip);
bus->irq = -1;
pci_disable_msi(chip->pci);
chip->msi = 0;
err = azx_acquire_irq(chip, 1);
if (err < 0)
return err;
return 0;
}
/* DMA page allocation helpers. */
static int dma_alloc_pages(struct hdac_bus *bus,
int type,
size_t size,
struct snd_dma_buffer *buf)
{
struct azx *chip = bus_to_azx(bus);
int err;
err = snd_dma_alloc_pages(type,
bus->dev,
size, buf);
if (err < 0)
return err;
mark_pages_wc(chip, buf, true);
return 0;
}
static void dma_free_pages(struct hdac_bus *bus, struct snd_dma_buffer *buf)
{
struct azx *chip = bus_to_azx(bus);
mark_pages_wc(chip, buf, false);
snd_dma_free_pages(buf);
}
static int substream_alloc_pages(struct azx *chip,
struct snd_pcm_substream *substream,
size_t size)
{
struct azx_dev *azx_dev = get_azx_dev(substream);
int ret;
mark_runtime_wc(chip, azx_dev, substream, false);
ret = snd_pcm_lib_malloc_pages(substream, size);
if (ret < 0)
return ret;
mark_runtime_wc(chip, azx_dev, substream, true);
return 0;
}
static int substream_free_pages(struct azx *chip,
struct snd_pcm_substream *substream)
{
struct azx_dev *azx_dev = get_azx_dev(substream);
mark_runtime_wc(chip, azx_dev, substream, false);
return snd_pcm_lib_free_pages(substream);
}
static void pcm_mmap_prepare(struct snd_pcm_substream *substream,
struct vm_area_struct *area)
{
#ifdef CONFIG_X86
struct azx_pcm *apcm = snd_pcm_substream_chip(substream);
struct azx *chip = apcm->chip;
if (!azx_snoop(chip) && chip->driver_type != AZX_DRIVER_CMEDIA)
area->vm_page_prot = pgprot_writecombine(area->vm_page_prot);
#endif
}
static const struct hdac_io_ops pci_hda_io_ops = {
.reg_writel = pci_azx_writel,
.reg_readl = pci_azx_readl,
.reg_writew = pci_azx_writew,
.reg_readw = pci_azx_readw,
.reg_writeb = pci_azx_writeb,
.reg_readb = pci_azx_readb,
.dma_alloc_pages = dma_alloc_pages,
.dma_free_pages = dma_free_pages,
};
static const struct hda_controller_ops pci_hda_ops = {
.disable_msi_reset_irq = disable_msi_reset_irq,
.substream_alloc_pages = substream_alloc_pages,
.substream_free_pages = substream_free_pages,
.pcm_mmap_prepare = pcm_mmap_prepare,
.position_check = azx_position_check,
.link_power = azx_intel_link_power,
};
static int azx_probe(struct pci_dev *pci,
const struct pci_device_id *pci_id)
{
static int dev;
struct snd_card *card;
struct hda_intel *hda;
struct azx *chip;
bool schedule_probe;
int err;
if (dev >= SNDRV_CARDS)
return -ENODEV;
if (!enable[dev]) {
dev++;
return -ENOENT;
}
err = snd_card_new(&pci->dev, index[dev], id[dev], THIS_MODULE,
0, &card);
if (err < 0) {
dev_err(&pci->dev, "Error creating card!\n");
return err;
}
err = azx_create(card, pci, dev, pci_id->driver_data, &chip);
if (err < 0)
goto out_free;
card->private_data = chip;
hda = container_of(chip, struct hda_intel, chip);
pci_set_drvdata(pci, card);
err = register_vga_switcheroo(chip);
if (err < 0) {
dev_err(card->dev, "Error registering vga_switcheroo client\n");
goto out_free;
}
if (check_hdmi_disabled(pci)) {
dev_info(card->dev, "VGA controller is disabled\n");
dev_info(card->dev, "Delaying initialization\n");
chip->disabled = true;
}
schedule_probe = !chip->disabled;
#ifdef CONFIG_SND_HDA_PATCH_LOADER
if (patch[dev] && *patch[dev]) {
dev_info(card->dev, "Applying patch firmware '%s'\n",
patch[dev]);
err = request_firmware_nowait(THIS_MODULE, true, patch[dev],
&pci->dev, GFP_KERNEL, card,
azx_firmware_cb);
if (err < 0)
goto out_free;
schedule_probe = false; /* continued in azx_firmware_cb() */
}
#endif /* CONFIG_SND_HDA_PATCH_LOADER */
#ifndef CONFIG_SND_HDA_I915
if (CONTROLLER_IN_GPU(pci))
dev_err(card->dev, "Haswell/Broadwell HDMI/DP must build in CONFIG_SND_HDA_I915\n");
#endif
if (schedule_probe)
schedule_work(&hda->probe_work);
dev++;
if (chip->disabled)
complete_all(&hda->probe_wait);
return 0;
out_free:
snd_card_free(card);
return err;
}
#ifdef CONFIG_PM
/* On some boards setting power_save to a non 0 value leads to clicking /
* popping sounds when ever we enter/leave powersaving mode. Ideally we would
* figure out how to avoid these sounds, but that is not always feasible.
* So we keep a list of devices where we disable powersaving as its known
* to causes problems on these devices.
*/
static struct snd_pci_quirk power_save_blacklist[] = {
/* https://bugzilla.redhat.com/show_bug.cgi?id=1525104 */
SND_PCI_QUIRK(0x1849, 0xc892, "Asrock B85M-ITX", 0),
/* https://bugzilla.redhat.com/show_bug.cgi?id=1525104 */
SND_PCI_QUIRK(0x1849, 0x7662, "Asrock H81M-HDS", 0),
/* https://bugzilla.redhat.com/show_bug.cgi?id=1525104 */
SND_PCI_QUIRK(0x1043, 0x8733, "Asus Prime X370-Pro", 0),
/* https://bugzilla.redhat.com/show_bug.cgi?id=1581607 */
SND_PCI_QUIRK(0x1558, 0x3501, "Clevo W35xSS_370SS", 0),
/* https://bugzilla.redhat.com/show_bug.cgi?id=1525104 */
/* Note the P55A-UD3 and Z87-D3HP share the subsys id for the HDA dev */
SND_PCI_QUIRK(0x1458, 0xa002, "Gigabyte P55A-UD3 / Z87-D3HP", 0),
/* https://bugzilla.kernel.org/show_bug.cgi?id=199607 */
SND_PCI_QUIRK(0x8086, 0x2057, "Intel NUC5i7RYB", 0),
/* https://bugzilla.redhat.com/show_bug.cgi?id=1520902 */
SND_PCI_QUIRK(0x8086, 0x2068, "Intel NUC7i3BNB", 0),
/* https://bugzilla.redhat.com/show_bug.cgi?id=1572975 */
SND_PCI_QUIRK(0x17aa, 0x36a7, "Lenovo C50 All in one", 0),
/* https://bugzilla.kernel.org/show_bug.cgi?id=198611 */
SND_PCI_QUIRK(0x17aa, 0x2227, "Lenovo X1 Carbon 3rd Gen", 0),
{}
};
#endif /* CONFIG_PM */
/* number of codec slots for each chipset: 0 = default slots (i.e. 4) */
static unsigned int azx_max_codecs[AZX_NUM_DRIVERS] = {
[AZX_DRIVER_NVIDIA] = 8,
[AZX_DRIVER_TERA] = 1,
};
static int azx_probe_continue(struct azx *chip)
{
struct hda_intel *hda = container_of(chip, struct hda_intel, chip);
struct hdac_bus *bus = azx_bus(chip);
struct pci_dev *pci = chip->pci;
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
struct hda_codec *codec;
int dev = chip->dev_index;
int val;
int err;
hda->probe_continued = 1;
/* bind with i915 if needed */
if (chip->driver_caps & AZX_DCAPS_I915_COMPONENT) {
err = snd_hdac_i915_init(bus);
if (err < 0) {
/* if the controller is bound only with HDMI/DP
* (for HSW and BDW), we need to abort the probe;
* for other chips, still continue probing as other
* codecs can be on the same link.
*/
if (CONTROLLER_IN_GPU(pci)) {
dev_err(chip->card->dev,
"HSW/BDW HD-audio HDMI/DP requires binding with gfx driver\n");
goto out_free;
} else {
/* don't bother any longer */
chip->driver_caps &=
~(AZX_DCAPS_I915_COMPONENT | AZX_DCAPS_I915_POWERWELL);
}
}
}
/* Request display power well for the HDA controller or codec. For
* Haswell/Broadwell, both the display HDA controller and codec need
* this power. For other platforms, like Baytrail/Braswell, only the
* display codec needs the power and it can be released after probe.
*/
if (chip->driver_caps & AZX_DCAPS_I915_POWERWELL) {
/* HSW/BDW controllers need this power */
if (CONTROLLER_IN_GPU(pci))
hda->need_i915_power = 1;
err = snd_hdac_display_power(bus, true);
if (err < 0) {
dev_err(chip->card->dev,
"Cannot turn on display power on i915\n");
goto i915_power_fail;
}
}
err = azx_first_init(chip);
if (err < 0)
goto out_free;
#ifdef CONFIG_SND_HDA_INPUT_BEEP
chip->beep_mode = beep_mode[dev];
#endif
/* create codec instances */
err = azx_probe_codecs(chip, azx_max_codecs[chip->driver_type]);
if (err < 0)
goto out_free;
#ifdef CONFIG_SND_HDA_PATCH_LOADER
if (chip->fw) {
err = snd_hda_load_patch(&chip->bus, chip->fw->size,
chip->fw->data);
if (err < 0)
goto out_free;
#ifndef CONFIG_PM
release_firmware(chip->fw); /* no longer needed */
chip->fw = NULL;
#endif
}
#endif
if ((probe_only[dev] & 1) == 0) {
err = azx_codec_configure(chip);
if (err < 0)
goto out_free;
}
err = snd_card_register(chip->card);
if (err < 0)
goto out_free;
chip->running = 1;
azx_add_card_list(chip);
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
val = power_save;
#ifdef CONFIG_PM
if (pm_blacklist) {
const struct snd_pci_quirk *q;
q = snd_pci_quirk_lookup(chip->pci, power_save_blacklist);
if (q && val) {
dev_info(chip->card->dev, "device %04x:%04x is on the power_save blacklist, forcing power_save to 0\n",
q->subvendor, q->subdevice);
val = 0;
}
}
#endif /* CONFIG_PM */
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
/*
* The discrete GPU cannot power down unless the HDA controller runtime
* suspends, so activate runtime PM on codecs even if power_save == 0.
*/
if (use_vga_switcheroo(hda))
list_for_each_codec(codec, &chip->bus)
codec->auto_runtime_pm = 1;
snd_hda_set_power_save(&chip->bus, val * 1000);
vga_switcheroo: Use device link for HDA controller Back in 2013, runtime PM for GPUs with integrated HDA controller was introduced with commits 0d69704ae348 ("gpu/vga_switcheroo: add driver control power feature. (v3)") and 246efa4a072f ("snd/hda: add runtime suspend/resume on optimus support (v4)"). Briefly, the idea was that the HDA controller is forced on and off in unison with the GPU. The original code is mostly still in place even though it was never a 100% perfect solution: E.g. on access to the HDA controller, the GPU is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there are no provisions to keep it resumed until access to the HDA controller has ceased: The GPU autosuspends after 5 seconds, rendering the HDA controller inaccessible. Additionally, a kludge is required when hda_intel.c probes: It has to check whether the GPU is powered down (check_hdmi_disabled()) and defer probing if so. However in the meantime (in v4.10) the driver core has gained a feature called device links which promises to solve such issues in a clean way: It allows us to declare a dependency from the HDA controller (consumer) to the GPU (supplier). The PM core then automagically ensures that the GPU is runtime resumed as long as the HDA controller's ->probe hook is executed and whenever the HDA controller is accessed. By default, the HDA controller has a dependency on its parent, a PCIe Root Port. Adding a device link creates another dependency on its sibling: PCIe Root Port ^ ^ | | | | HDA ===> GPU The device link is not only used for runtime PM, it also guarantees that on system sleep, the HDA controller suspends before the GPU and resumes after the GPU, and on system shutdown the HDA controller's ->shutdown hook is executed before the one of the GPU. It is a complete solution. Using this functionality is as simple as calling device_link_add(), which results in a dmesg entry like this: pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0 The code for the GPU-governed audio power management can thus be removed (except where it's still needed for legacy manual power control). The device link is added in a PCI quirk rather than in hda_intel.c. It is therefore legal for the GPU to runtime suspend to D3cold even if the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL is not enabled, for accesses to the HDA controller will cause the GPU to wake up regardless if they're occurring outside of hda_intel.c (think config space readout via sysfs). Contrary to the previous implementation, the HDA controller's power state is now self-governed, rather than GPU-governed, whereas the GPU's power state is no longer fully self-governed. (The HDA controller needs to runtime suspend before the GPU can.) It is thus crucial that runtime PM is always activated on the HDA controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which is the default), lest the GPU stays awake. This is achieved by setting the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME flag on the HDA controller. A side effect is that power consumption might be reduced if the GPU is in use but the HDA controller is not, because the HDA controller is now allowed to go to D3hot. Before, it was forced to stay in D0 as long as the GPU was in use. (There is no reduction in power consumption on my Nvidia GK107, but there might be on other chips.) The code paths for legacy manual power control are adjusted such that runtime PM is disabled during power off, thereby preventing the PM core from resuming the HDA controller. Note that the device link is not only added on vga_switcheroo capable systems, but for *any* GPU with integrated HDA controller. The idea is that the HDA controller streams audio via connectors located on the GPU, so the GPU needs to be on for the HDA controller to do anything useful. This commit implicitly fixes an unbalanced runtime PM ref upon unbind of hda_intel.c: On ->probe, a runtime PM ref was previously released under the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but on ->remove a runtime PM ref was only acquired under the first of those conditions. Thus, binding and unbinding the driver twice on a vga_switcheroo capable system caused the runtime PM refcount to drop below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag is now always set if use_vga_switcheroo is true. For more information on device links please refer to: https://www.kernel.org/doc/html/latest/driver-api/device_link.html Documentation/driver-api/device_link.rst Cc: Dave Airlie <airlied@redhat.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Peter Wu <peter@lekensteyn.nl> Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro Signed-off-by: Lukas Wunner <lukas@wunner.de> Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
2018-03-03 09:53:24 +00:00
if (azx_has_pm_runtime(chip))
ALSA: hda - Autosuspend controller after probe even if codecs are already suspended azx_probe_continue() uses pm_runtime_put_noidle() to drop the rpm usage_count, which means that if it's the last reference the autosuspend of the controller won't actually happen. So if the codecs autosuspend before the azx_probe_continue() drops the last reference we'll fail to autosuspend the controller. This does happen in practice, but not every time. As can be seen in [1] the controller autosuspend attempt fails due to the usage_count when suspending the codecs. A bit later we see the the contoller usage_count dropping to zero without further attempts at autosuspend. Fix the problem by using pm_runtime_put_autosuspend() instead, which will kick off the autosuspend of the controller even if the codecs are already asleep. As can be seen in [2] the controller autosuspend still fails while suspending the codecs, but later on we see another autosuspend attempt after dropping the usage_count to 0. I was also a bit worried that there might still be a race between the controller autosuspend and the rest of the code in azx_probe_continue(). So I also tried replacing the the put_noidle() with put_sync_suspend(). No explosions occurred, so I'm somewhat satisfied that there are no serious problems in this area. [1] kworker/1:2-122 [001] .... 63.661310: __pm_runtime_suspend: hdaudioC0D0 usage_count 0 kworker/1:2-122 [001] d..2 63.661316: rpm_suspend: hdaudioC0D0 flags-d cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/1:2-122 [001] d..1 63.661317: rpm_check_suspend_allowed: hdaudioC0D0 retval 0 kworker/1:2-122 [001] d..2 63.661332: rpm_return_int: rpm_suspend+0x406/0x5e8:hdaudioC0D0 ret=0 kworker/1:1-72 [001] d..2 63.661543: rpm_suspend: hdaudioC0D0 flags-a cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/1:1-72 [001] d..1 63.661544: rpm_check_suspend_allowed: hdaudioC0D0 retval 0 kworker/1:1-72 [001] .... 63.661545: hda_codec_runtime_suspend: hdaudioC0D0 suspend kworker/1:1-72 [001] d..2 63.661614: rpm_idle: 0000:00:03.0 flags-1 cnt-1 dep-0 auto-1 p-0 irq-0 child-0 kworker/1:1-72 [001] d..1 63.661615: rpm_check_suspend_allowed: 0000:00:03.0 usage_count 1 kworker/1:1-72 [001] d..1 63.661615: rpm_check_suspend_allowed: 0000:00:03.0 retval -11 kworker/1:1-72 [001] d..2 63.661616: rpm_return_int: rpm_idle+0x249/0x487:0000:00:03.0 ret=-11 kworker/1:1-72 [001] d..2 63.661616: rpm_return_int: rpm_suspend+0x406/0x5e8:hdaudioC0D0 ret=0 kworker/1:2-122 [001] d..2 63.664834: rpm_idle: hdaudioC0D0 flags-8 cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/1:2-122 [001] d..1 63.664835: rpm_check_suspend_allowed: hdaudioC0D0 retval 1 kworker/1:2-122 [001] d..2 63.664836: rpm_return_int: rpm_idle+0x249/0x487:hdaudioC0D0 ret=-11 kworker/1:2-122 [001] d..2 63.664841: rpm_idle: hdaudioC0D0 flags-8 cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/1:2-122 [001] d..1 63.664841: rpm_check_suspend_allowed: hdaudioC0D0 retval 1 kworker/1:2-122 [001] d..2 63.664841: rpm_return_int: rpm_idle+0x249/0x487:hdaudioC0D0 ret=-11 kworker/1:2-122 [001] .... 63.664842: azx_probe_continue: 0000:00:03.0 usage_count=0 [2] kworker/0:0-4 [000] .... 50.354567: __pm_runtime_suspend: hdaudioC0D0 usage_count 0 kworker/0:0-4 [000] d..2 50.354574: rpm_suspend: hdaudioC0D0 flags-d cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/0:0-4 [000] d..1 50.354575: rpm_check_suspend_allowed: hdaudioC0D0 retval 0 kworker/0:0-4 [000] d..2 50.354589: rpm_return_int: rpm_suspend+0x406/0x5e8:hdaudioC0D0 ret=0 kworker/0:2-135 [000] d..2 50.354809: rpm_suspend: hdaudioC0D0 flags-a cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/0:2-135 [000] d..1 50.354810: rpm_check_suspend_allowed: hdaudioC0D0 retval 0 kworker/0:2-135 [000] .... 50.354816: hda_codec_runtime_suspend: hdaudioC0D0 suspend kworker/0:2-135 [000] d..2 50.354908: rpm_idle: 0000:00:03.0 flags-1 cnt-1 dep-0 auto-1 p-0 irq-0 child-0 kworker/0:2-135 [000] d..1 50.354909: rpm_check_suspend_allowed: 0000:00:03.0 usage_count 1 kworker/0:2-135 [000] d..1 50.354909: rpm_check_suspend_allowed: 0000:00:03.0 retval -11 kworker/0:2-135 [000] d..2 50.354909: rpm_return_int: rpm_idle+0x249/0x487:0000:00:03.0 ret=-11 kworker/0:2-135 [000] d..2 50.354910: rpm_return_int: rpm_suspend+0x406/0x5e8:hdaudioC0D0 ret=0 kworker/0:0-4 [000] d..2 50.373791: rpm_idle: hdaudioC0D0 flags-8 cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/0:0-4 [000] d..1 50.373792: rpm_check_suspend_allowed: hdaudioC0D0 retval 1 kworker/0:0-4 [000] d..2 50.373793: rpm_return_int: rpm_idle+0x249/0x487:hdaudioC0D0 ret=-11 kworker/0:0-4 [000] d..2 50.373797: rpm_idle: hdaudioC0D0 flags-8 cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/0:0-4 [000] d..1 50.373798: rpm_check_suspend_allowed: hdaudioC0D0 retval 1 kworker/0:0-4 [000] d..2 50.373798: rpm_return_int: rpm_idle+0x249/0x487:hdaudioC0D0 ret=-11 kworker/0:0-4 [000] .... 50.373799: __pm_runtime_suspend: 0000:00:03.0 usage_count 0 kworker/0:0-4 [000] d..2 50.373800: rpm_suspend: 0000:00:03.0 flags-d cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/0:0-4 [000] d..1 50.373800: rpm_check_suspend_allowed: 0000:00:03.0 retval 0 kworker/0:0-4 [000] d..2 50.373803: rpm_return_int: rpm_suspend+0x406/0x5e8:0000:00:03.0 ret=0 kworker/0:0-4 [000] d..2 50.385164: rpm_suspend: 0000:00:03.0 flags-a cnt-0 dep-0 auto-1 p-0 irq-0 child-0 kworker/0:0-4 [000] d..1 50.385165: rpm_check_suspend_allowed: 0000:00:03.0 retval 0 kworker/0:0-4 [000] .... 50.385174: azx_runtime_suspend: 0000:00:03.0 azx suspend releaseing power well kworker/0:0-4 [000] .... 50.385179: azx_runtime_suspend: 0000:00:03.0 azx suspend kworker/0:0-4 [000] d..2 50.386872: rpm_return_int: rpm_suspend+0x406/0x5e8:0000:00:03.0 ret=0 Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2016-02-26 17:39:57 +00:00
pm_runtime_put_autosuspend(&pci->dev);
out_free:
if ((chip->driver_caps & AZX_DCAPS_I915_POWERWELL)
&& !hda->need_i915_power)
snd_hdac_display_power(bus, false);
i915_power_fail:
if (err < 0)
hda->init_failed = 1;
complete_all(&hda->probe_wait);
return err;
}
static void azx_remove(struct pci_dev *pci)
{
struct snd_card *card = pci_get_drvdata(pci);
struct azx *chip;
struct hda_intel *hda;
if (card) {
/* cancel the pending probing work */
chip = card->private_data;
hda = container_of(chip, struct hda_intel, chip);
2017-01-02 10:37:04 +00:00
/* FIXME: below is an ugly workaround.
* Both device_release_driver() and driver_probe_device()
* take *both* the device's and its parent's lock before
* calling the remove() and probe() callbacks. The codec
* probe takes the locks of both the codec itself and its
* parent, i.e. the PCI controller dev. Meanwhile, when
* the PCI controller is unbound, it takes its lock, too
* ==> ouch, a deadlock!
* As a workaround, we unlock temporarily here the controller
* device during cancel_work_sync() call.
*/
device_unlock(&pci->dev);
cancel_work_sync(&hda->probe_work);
2017-01-02 10:37:04 +00:00
device_lock(&pci->dev);
ALSA: hda - add runtime PM support Runtime PM can bring more power saving: - When the controller is suspended, its parent device will also have a chance to suspend. - PCI subsystem can choose the lowest power state the controller can signal wake up from. This state can be D3cold on platforms with ACPI PM support. And runtime PM can provide a gerneral sysfs interface for a system policy manager. Runtime PM support is based on current HDA power saving implementation. The user can enable runtime PM on platfroms that provide acceptable latency on transition from D3 to D0. Details: - When both power saving and runtime PM are enabled: -- If a codec supports 'stop-clock' in D3, it will request suspending the controller after it enters D3 and request resuming the controller before back to D0. Thus the controller will be suspended only when all codecs are suspended and support stop-clock in D3. -- User IO operations and HW wakeup signal can resume the controller back to D0. - If runtime PM is disabled, power saving just works as before. - If power saving is disabled, the controller won't be suspended because the power usage counter can never be 0. More about 'stop-clock' feature: If a codec can support targeted pass-through operations in D3 state when there is no BCLK present on the link, it will set CLKSTOP flag in the supported power states and report PS-ClkStopOk when entering D3 state. Please refer to HDA spec section 7.3.3.10 Power state and 7.3.4.12 Supported Power State. [Fixed CONFIG_PM_RUNTIME dependency in hda_intel.c by tiwai] Signed-off-by: Mengdong Lin <mengdong.lin@intel.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-08-23 09:32:30 +00:00
snd_card_free(card);
}
}
static void azx_shutdown(struct pci_dev *pci)
{
struct snd_card *card = pci_get_drvdata(pci);
struct azx *chip;
if (!card)
return;
chip = card->private_data;
if (chip && chip->running)
azx_stop_chip(chip);
}
/* PCI IDs */
static const struct pci_device_id azx_ids[] = {
/* CPT */
{ PCI_DEVICE(0x8086, 0x1c20),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH_NOPM },
/* PBG */
{ PCI_DEVICE(0x8086, 0x1d20),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH_NOPM },
/* Panther Point */
{ PCI_DEVICE(0x8086, 0x1e20),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH_NOPM },
/* Lynx Point */
{ PCI_DEVICE(0x8086, 0x8c20),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH },
/* 9 Series */
{ PCI_DEVICE(0x8086, 0x8ca0),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH },
/* Wellsburg */
{ PCI_DEVICE(0x8086, 0x8d20),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH },
{ PCI_DEVICE(0x8086, 0x8d21),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH },
/* Lewisburg */
{ PCI_DEVICE(0x8086, 0xa1f0),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_SKYLAKE },
{ PCI_DEVICE(0x8086, 0xa270),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_SKYLAKE },
/* Lynx Point-LP */
{ PCI_DEVICE(0x8086, 0x9c20),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH },
/* Lynx Point-LP */
{ PCI_DEVICE(0x8086, 0x9c21),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH },
/* Wildcat Point-LP */
{ PCI_DEVICE(0x8086, 0x9ca0),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_PCH },
/* Sunrise Point */
{ PCI_DEVICE(0x8086, 0xa170),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE },
/* Sunrise Point-LP */
{ PCI_DEVICE(0x8086, 0x9d70),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE },
/* Kabylake */
{ PCI_DEVICE(0x8086, 0xa171),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE },
/* Kabylake-LP */
{ PCI_DEVICE(0x8086, 0x9d71),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE },
/* Kabylake-H */
{ PCI_DEVICE(0x8086, 0xa2f0),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE },
/* Coffelake */
{ PCI_DEVICE(0x8086, 0xa348),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE},
/* Cannonlake */
{ PCI_DEVICE(0x8086, 0x9dc8),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE},
/* Icelake */
{ PCI_DEVICE(0x8086, 0x34c8),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_SKYLAKE},
/* Broxton-P(Apollolake) */
{ PCI_DEVICE(0x8086, 0x5a98),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_BROXTON },
/* Broxton-T */
{ PCI_DEVICE(0x8086, 0x1a98),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_BROXTON },
/* Gemini-Lake */
{ PCI_DEVICE(0x8086, 0x3198),
.driver_data = AZX_DRIVER_SKL | AZX_DCAPS_INTEL_BROXTON },
/* Haswell */
{ PCI_DEVICE(0x8086, 0x0a0c),
.driver_data = AZX_DRIVER_HDMI | AZX_DCAPS_INTEL_HASWELL },
{ PCI_DEVICE(0x8086, 0x0c0c),
.driver_data = AZX_DRIVER_HDMI | AZX_DCAPS_INTEL_HASWELL },
{ PCI_DEVICE(0x8086, 0x0d0c),
.driver_data = AZX_DRIVER_HDMI | AZX_DCAPS_INTEL_HASWELL },
/* Broadwell */
{ PCI_DEVICE(0x8086, 0x160c),
.driver_data = AZX_DRIVER_HDMI | AZX_DCAPS_INTEL_BROADWELL },
/* 5 Series/3400 */
{ PCI_DEVICE(0x8086, 0x3b56),
.driver_data = AZX_DRIVER_SCH | AZX_DCAPS_INTEL_PCH_NOPM },
/* Poulsbo */
{ PCI_DEVICE(0x8086, 0x811b),
.driver_data = AZX_DRIVER_SCH | AZX_DCAPS_INTEL_PCH_BASE },
/* Oaktrail */
{ PCI_DEVICE(0x8086, 0x080a),
.driver_data = AZX_DRIVER_SCH | AZX_DCAPS_INTEL_PCH_BASE },
/* BayTrail */
{ PCI_DEVICE(0x8086, 0x0f04),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_BAYTRAIL },
/* Braswell */
{ PCI_DEVICE(0x8086, 0x2284),
.driver_data = AZX_DRIVER_PCH | AZX_DCAPS_INTEL_BRASWELL },
/* ICH6 */
{ PCI_DEVICE(0x8086, 0x2668),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* ICH7 */
{ PCI_DEVICE(0x8086, 0x27d8),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* ESB2 */
{ PCI_DEVICE(0x8086, 0x269a),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* ICH8 */
{ PCI_DEVICE(0x8086, 0x284b),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* ICH9 */
{ PCI_DEVICE(0x8086, 0x293e),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* ICH9 */
{ PCI_DEVICE(0x8086, 0x293f),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* ICH10 */
{ PCI_DEVICE(0x8086, 0x3a3e),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* ICH10 */
{ PCI_DEVICE(0x8086, 0x3a6e),
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_INTEL_ICH },
/* Generic Intel */
{ PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_ANY_ID),
.class = PCI_CLASS_MULTIMEDIA_HD_AUDIO << 8,
.class_mask = 0xffffff,
.driver_data = AZX_DRIVER_ICH | AZX_DCAPS_NO_ALIGN_BUFSIZE },
/* ATI SB 450/600/700/800/900 */
{ PCI_DEVICE(0x1002, 0x437b),
.driver_data = AZX_DRIVER_ATI | AZX_DCAPS_PRESET_ATI_SB },
{ PCI_DEVICE(0x1002, 0x4383),
.driver_data = AZX_DRIVER_ATI | AZX_DCAPS_PRESET_ATI_SB },
/* AMD Hudson */
{ PCI_DEVICE(0x1022, 0x780d),
.driver_data = AZX_DRIVER_GENERIC | AZX_DCAPS_PRESET_ATI_SB },
/* AMD Raven */
{ PCI_DEVICE(0x1022, 0x15e3),
.driver_data = AZX_DRIVER_GENERIC | AZX_DCAPS_PRESET_ATI_SB |
AZX_DCAPS_PM_RUNTIME },
/* ATI HDMI */
{ PCI_DEVICE(0x1002, 0x0002),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0x1308),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0x157a),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0x15b3),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0x793b),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0x7919),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0x960f),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0x970f),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0x9840),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaa00),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa08),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa10),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa18),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa20),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa28),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa30),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa38),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa40),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa48),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa50),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa58),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa60),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa68),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa80),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa88),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa90),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0xaa98),
.driver_data = AZX_DRIVER_ATIHDMI | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(0x1002, 0x9902),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaaa0),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaaa8),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaab0),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaac0),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaac8),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaad8),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaae8),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaae0),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
{ PCI_DEVICE(0x1002, 0xaaf0),
.driver_data = AZX_DRIVER_ATIHDMI_NS | AZX_DCAPS_PRESET_ATI_HDMI_NS },
/* VIA VT8251/VT8237A */
{ PCI_DEVICE(0x1106, 0x3288), .driver_data = AZX_DRIVER_VIA },
/* VIA GFX VT7122/VX900 */
{ PCI_DEVICE(0x1106, 0x9170), .driver_data = AZX_DRIVER_GENERIC },
/* VIA GFX VT6122/VX11 */
{ PCI_DEVICE(0x1106, 0x9140), .driver_data = AZX_DRIVER_GENERIC },
/* SIS966 */
{ PCI_DEVICE(0x1039, 0x7502), .driver_data = AZX_DRIVER_SIS },
/* ULI M5461 */
{ PCI_DEVICE(0x10b9, 0x5461), .driver_data = AZX_DRIVER_ULI },
/* NVIDIA MCP */
{ PCI_DEVICE(PCI_VENDOR_ID_NVIDIA, PCI_ANY_ID),
.class = PCI_CLASS_MULTIMEDIA_HD_AUDIO << 8,
.class_mask = 0xffffff,
.driver_data = AZX_DRIVER_NVIDIA | AZX_DCAPS_PRESET_NVIDIA },
/* Teradici */
{ PCI_DEVICE(0x6549, 0x1200),
.driver_data = AZX_DRIVER_TERA | AZX_DCAPS_NO_64BIT },
{ PCI_DEVICE(0x6549, 0x2200),
.driver_data = AZX_DRIVER_TERA | AZX_DCAPS_NO_64BIT },
/* Creative X-Fi (CA0110-IBG) */
/* CTHDA chips */
{ PCI_DEVICE(0x1102, 0x0010),
.driver_data = AZX_DRIVER_CTHDA | AZX_DCAPS_PRESET_CTHDA },
{ PCI_DEVICE(0x1102, 0x0012),
.driver_data = AZX_DRIVER_CTHDA | AZX_DCAPS_PRESET_CTHDA },
#if !IS_ENABLED(CONFIG_SND_CTXFI)
/* the following entry conflicts with snd-ctxfi driver,
* as ctxfi driver mutates from HD-audio to native mode with
* a special command sequence.
*/
{ PCI_DEVICE(PCI_VENDOR_ID_CREATIVE, PCI_ANY_ID),
.class = PCI_CLASS_MULTIMEDIA_HD_AUDIO << 8,
.class_mask = 0xffffff,
.driver_data = AZX_DRIVER_CTX | AZX_DCAPS_CTX_WORKAROUND |
AZX_DCAPS_NO_64BIT | AZX_DCAPS_POSFIX_LPIB },
#else
/* this entry seems still valid -- i.e. without emu20kx chip */
{ PCI_DEVICE(0x1102, 0x0009),
.driver_data = AZX_DRIVER_CTX | AZX_DCAPS_CTX_WORKAROUND |
AZX_DCAPS_NO_64BIT | AZX_DCAPS_POSFIX_LPIB },
#endif
/* CM8888 */
{ PCI_DEVICE(0x13f6, 0x5011),
.driver_data = AZX_DRIVER_CMEDIA |
AZX_DCAPS_NO_MSI | AZX_DCAPS_POSFIX_LPIB | AZX_DCAPS_SNOOP_OFF },
/* Vortex86MX */
{ PCI_DEVICE(0x17f3, 0x3010), .driver_data = AZX_DRIVER_GENERIC },
/* VMware HDAudio */
{ PCI_DEVICE(0x15ad, 0x1977), .driver_data = AZX_DRIVER_GENERIC },
/* AMD/ATI Generic, PCI class code and Vendor ID for HD Audio */
{ PCI_DEVICE(PCI_VENDOR_ID_ATI, PCI_ANY_ID),
.class = PCI_CLASS_MULTIMEDIA_HD_AUDIO << 8,
.class_mask = 0xffffff,
.driver_data = AZX_DRIVER_GENERIC | AZX_DCAPS_PRESET_ATI_HDMI },
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_ANY_ID),
.class = PCI_CLASS_MULTIMEDIA_HD_AUDIO << 8,
.class_mask = 0xffffff,
.driver_data = AZX_DRIVER_GENERIC | AZX_DCAPS_PRESET_ATI_HDMI },
{ 0, }
};
MODULE_DEVICE_TABLE(pci, azx_ids);
/* pci_driver definition */
static struct pci_driver azx_driver = {
.name = KBUILD_MODNAME,
.id_table = azx_ids,
.probe = azx_probe,
.remove = azx_remove,
.shutdown = azx_shutdown,
.driver = {
.pm = AZX_PM_OPS,
},
};
module_pci_driver(azx_driver);