linux/drivers/pci/bus.c

468 lines
11 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* From setup-res.c, by:
* Dave Rusling (david.rusling@reo.mts.dec.com)
* David Mosberger (davidm@cs.arizona.edu)
* David Miller (davem@redhat.com)
* Ivan Kokshaysky (ink@jurassic.park.msu.ru)
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/pci.h>
#include <linux/errno.h>
#include <linux/ioport.h>
PCI: move OF status = "disabled" detection to dev->match_driver The blamed commit has broken probing on arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi when &enetc_port0 (PCI function 0) has status = "disabled". Background: pci_scan_slot() has logic to say that if the function 0 of a device is absent, the entire device is absent and we can skip the other functions entirely. Traditionally, this has meant that pci_bus_read_dev_vendor_id() returns an error code for that function. However, since the blamed commit, there is an extra confounding condition: function 0 of the device exists and has a valid vendor id, but it is disabled in the device tree. In that case, pci_scan_slot() would incorrectly skip the entire device instead of just that function. In the case of NXP LS1028A, status = "disabled" does not mean that the PCI function's config space is not available for reading. It is, but the Ethernet port is just not functionally useful with a particular SerDes protocol configuration (0x9999) due to pinmuxing constraints of the Soc. So, pci_scan_slot() skips all other functions on the ENETC ECAM (enetc_port1, enetc_port2, enetc_mdio_pf3 etc) when just enetc_port0 had to not be probed. There is an additional regression introduced by the change, caused by its fundamental premise. The enetc driver needs to run code for all PCI functions, regardless of whether they're enabled or not in the device tree. That is no longer possible if the driver's probe function is no longer called. But Rob recommends that we move the of_device_is_available() detection to dev->match_driver, and this makes the PCI fixups still run on all functions, while just probing drivers for those functions that are enabled. So, a separate change in the enetc driver will have to move the workarounds to a PCI fixup. Fixes: 6fffbc7ae137 ("PCI: Honor firmware's device disabled status") Link: https://lore.kernel.org/netdev/CAL_JsqLsVYiPLx2kcHkDQ4t=hQVCR7NHziDwi9cCFUFhx48Qow@mail.gmail.com/ Suggested-by: Rob Herring <robh@kernel.org> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2023-08-03 13:58:56 +00:00
#include <linux/of.h>
#include <linux/of_platform.h>
#include <linux/proc_fs.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include "pci.h"
void pci_add_resource_offset(struct list_head *resources, struct resource *res,
resource_size_t offset)
{
struct resource_entry *entry;
entry = resource_list_create_entry(res, 0);
if (!entry) {
pr_err("PCI: can't add host bridge window %pR\n", res);
return;
}
entry->offset = offset;
resource_list_add_tail(entry, resources);
}
EXPORT_SYMBOL(pci_add_resource_offset);
void pci_add_resource(struct list_head *resources, struct resource *res)
{
pci_add_resource_offset(resources, res, 0);
}
EXPORT_SYMBOL(pci_add_resource);
void pci_free_resource_list(struct list_head *resources)
{
resource_list_free(resources);
}
EXPORT_SYMBOL(pci_free_resource_list);
void pci_bus_add_resource(struct pci_bus *bus, struct resource *res,
unsigned int flags)
{
struct pci_bus_resource *bus_res;
bus_res = kzalloc(sizeof(struct pci_bus_resource), GFP_KERNEL);
if (!bus_res) {
dev_err(&bus->dev, "can't add %pR resource\n", res);
return;
}
bus_res->res = res;
bus_res->flags = flags;
list_add_tail(&bus_res->list, &bus->resources);
}
struct resource *pci_bus_resource_n(const struct pci_bus *bus, int n)
{
struct pci_bus_resource *bus_res;
if (n < PCI_BRIDGE_RESOURCE_NUM)
return bus->resource[n];
n -= PCI_BRIDGE_RESOURCE_NUM;
list_for_each_entry(bus_res, &bus->resources, list) {
if (n-- == 0)
return bus_res->res;
}
return NULL;
}
EXPORT_SYMBOL_GPL(pci_bus_resource_n);
PCI: s390: Fix use-after-free of PCI resources with per-function hotplug On s390 PCI functions may be hotplugged individually even when they belong to a multi-function device. In particular on an SR-IOV device VFs may be removed and later re-added. In commit a50297cf8235 ("s390/pci: separate zbus creation from scanning") it was missed however that struct pci_bus and struct zpci_bus's resource list retained a reference to the PCI functions MMIO resources even though those resources are released and freed on hot-unplug. These stale resources may subsequently be claimed when the PCI function re-appears resulting in use-after-free. One idea of fixing this use-after-free in s390 specific code that was investigated was to simply keep resources around from the moment a PCI function first appeared until the whole virtual PCI bus created for a multi-function device disappears. The problem with this however is that due to the requirement of artificial MMIO addreesses (address cookies) extra logic is then needed to keep the address cookies compatible on re-plug. At the same time the MMIO resources semantically belong to the PCI function so tying their lifecycle to the function seems more logical. Instead a simpler approach is to remove the resources of an individually hot-unplugged PCI function from the PCI bus's resource list while keeping the resources of other PCI functions on the PCI bus untouched. This is done by introducing pci_bus_remove_resource() to remove an individual resource. Similarly the resource also needs to be removed from the struct zpci_bus's resource list. It turns out however, that there is really no need to add the MMIO resources to the struct zpci_bus's resource list at all and instead we can simply use the zpci_bar_struct's resource pointer directly. Fixes: a50297cf8235 ("s390/pci: separate zbus creation from scanning") Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com> Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com> Acked-by: Bjorn Helgaas <bhelgaas@google.com> Link: https://lore.kernel.org/r/20230306151014.60913-2-schnelle@linux.ibm.com Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
2023-03-06 15:10:11 +00:00
void pci_bus_remove_resource(struct pci_bus *bus, struct resource *res)
{
struct pci_bus_resource *bus_res, *tmp;
int i;
for (i = 0; i < PCI_BRIDGE_RESOURCE_NUM; i++) {
if (bus->resource[i] == res) {
bus->resource[i] = NULL;
return;
}
}
list_for_each_entry_safe(bus_res, tmp, &bus->resources, list) {
if (bus_res->res == res) {
list_del(&bus_res->list);
kfree(bus_res);
return;
}
}
}
void pci_bus_remove_resources(struct pci_bus *bus)
{
int i;
struct pci_bus_resource *bus_res, *tmp;
for (i = 0; i < PCI_BRIDGE_RESOURCE_NUM; i++)
bus->resource[i] = NULL;
list_for_each_entry_safe(bus_res, tmp, &bus->resources, list) {
list_del(&bus_res->list);
kfree(bus_res);
}
}
int devm_request_pci_bus_resources(struct device *dev,
struct list_head *resources)
{
struct resource_entry *win;
struct resource *parent, *res;
int err;
resource_list_for_each_entry(win, resources) {
res = win->res;
switch (resource_type(res)) {
case IORESOURCE_IO:
parent = &ioport_resource;
break;
case IORESOURCE_MEM:
parent = &iomem_resource;
break;
default:
continue;
}
err = devm_request_resource(dev, parent, res);
if (err)
return err;
}
return 0;
}
EXPORT_SYMBOL_GPL(devm_request_pci_bus_resources);
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
static struct pci_bus_region pci_32_bit = {0, 0xffffffffULL};
#ifdef CONFIG_ARCH_DMA_ADDR_T_64BIT
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
static struct pci_bus_region pci_64_bit = {0,
PCI: Add pci_bus_addr_t David Ahern reported that d63e2e1f3df9 ("sparc/PCI: Clip bridge windows to fit in upstream windows") fails to boot on sparc/T5-8: pci 0000:06:00.0: reg 0x184: can't handle BAR above 4GB (bus address 0x110204000) The problem is that sparc64 assumed that dma_addr_t only needed to hold DMA addresses, i.e., bus addresses returned via the DMA API (dma_map_single(), etc.), while the PCI core assumed dma_addr_t could hold *any* bus address, including raw BAR values. On sparc64, all DMA addresses fit in 32 bits, so dma_addr_t is a 32-bit type. However, BAR values can be 64 bits wide, so they don't fit in a dma_addr_t. d63e2e1f3df9 added new checking that tripped over this mismatch. Add pci_bus_addr_t, which is wide enough to hold any PCI bus address, including both raw BAR values and DMA addresses. This will be 64 bits on 64-bit platforms and on platforms with a 64-bit dma_addr_t. Then dma_addr_t only needs to be wide enough to hold addresses from the DMA API. [bhelgaas: changelog, bugzilla, Kconfig to ensure pci_bus_addr_t is at least as wide as dma_addr_t, documentation] Fixes: d63e2e1f3df9 ("sparc/PCI: Clip bridge windows to fit in upstream windows") Fixes: 23b13bc76f35 ("PCI: Fail safely if we can't handle BARs larger than 4GB") Link: http://lkml.kernel.org/r/CAE9FiQU1gJY1LYrxs+ma5LCTEEe4xmtjRG0aXJ9K_Tsu+m9Wuw@mail.gmail.com Link: http://lkml.kernel.org/r/1427857069-6789-1-git-send-email-yinghai@kernel.org Link: https://bugzilla.kernel.org/show_bug.cgi?id=96231 Reported-by: David Ahern <david.ahern@oracle.com> Tested-by: David Ahern <david.ahern@oracle.com> Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: David S. Miller <davem@davemloft.net> CC: stable@vger.kernel.org # v3.19+
2015-05-28 00:23:51 +00:00
(pci_bus_addr_t) 0xffffffffffffffffULL};
static struct pci_bus_region pci_high = {(pci_bus_addr_t) 0x100000000ULL,
(pci_bus_addr_t) 0xffffffffffffffffULL};
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
#endif
/*
* @res contains CPU addresses. Clip it so the corresponding bus addresses
* on @bus are entirely within @region. This is used to control the bus
* addresses of resources we allocate, e.g., we may need a resource that
* can be mapped by a 32-bit BAR.
*/
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
static void pci_clip_resource_to_region(struct pci_bus *bus,
struct resource *res,
struct pci_bus_region *region)
{
struct pci_bus_region r;
pcibios_resource_to_bus(bus, &r, res);
if (r.start < region->start)
r.start = region->start;
if (r.end > region->end)
r.end = region->end;
if (r.end < r.start)
res->end = res->start - 1;
else
pcibios_bus_to_resource(bus, res, &r);
}
static int pci_bus_alloc_from_region(struct pci_bus *bus, struct resource *res,
resource_size_t size, resource_size_t align,
resource_size_t min, unsigned long type_mask,
resource_alignf alignf,
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
void *alignf_data,
struct pci_bus_region *region)
{
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
struct resource *r, avail;
resource_size_t max;
int ret;
type_mask |= IORESOURCE_TYPE_BITS;
pci_bus_for_each_resource(bus, r) {
resource_size_t min_used = min;
if (!r)
continue;
/* type_mask must match */
if ((res->flags ^ r->flags) & type_mask)
continue;
/* We cannot allocate a non-prefetching resource
from a pre-fetching area */
if ((r->flags & IORESOURCE_PREFETCH) &&
!(res->flags & IORESOURCE_PREFETCH))
continue;
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
avail = *r;
pci_clip_resource_to_region(bus, &avail, region);
/*
* "min" is typically PCIBIOS_MIN_IO or PCIBIOS_MIN_MEM to
* protect badly documented motherboard resources, but if
* this is an already-configured bridge window, its start
* overrides "min".
*/
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
if (avail.start)
min_used = avail.start;
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
max = avail.end;
/* Don't bother if available space isn't large enough */
if (size > max - min_used + 1)
continue;
/* Ok, try it out.. */
ret = allocate_resource(r, res, size, min_used, max,
align, alignf, alignf_data);
if (ret == 0)
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
return 0;
}
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
return -ENOMEM;
}
/**
* pci_bus_alloc_resource - allocate a resource from a parent bus
* @bus: PCI bus
* @res: resource to allocate
* @size: size of resource to allocate
* @align: alignment of resource to allocate
* @min: minimum /proc/iomem address to allocate
* @type_mask: IORESOURCE_* type flags
* @alignf: resource alignment function
* @alignf_data: data argument for resource alignment function
*
* Given the PCI bus a device resides on, the size, minimum address,
* alignment and type, try to find an acceptable resource allocation
* for a specific device resource.
*/
int pci_bus_alloc_resource(struct pci_bus *bus, struct resource *res,
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
resource_size_t size, resource_size_t align,
resource_size_t min, unsigned long type_mask,
resource_alignf alignf,
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
void *alignf_data)
{
#ifdef CONFIG_ARCH_DMA_ADDR_T_64BIT
int rc;
if (res->flags & IORESOURCE_MEM_64) {
rc = pci_bus_alloc_from_region(bus, res, size, align, min,
type_mask, alignf, alignf_data,
&pci_high);
if (rc == 0)
return 0;
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
return pci_bus_alloc_from_region(bus, res, size, align, min,
type_mask, alignf, alignf_data,
&pci_64_bit);
}
PCI: Enforce bus address limits in resource allocation When allocating space for 32-bit BARs, we previously limited RESOURCE addresses so they would fit in 32 bits. However, the BUS address need not be the same as the resource address, and it's the bus address that must fit in the 32-bit BAR. This patch adds: - pci_clip_resource_to_region(), which clips a resource so it contains only the range that maps to the specified bus address region, e.g., to clip a resource to 32-bit bus addresses, and - pci_bus_alloc_from_region(), which allocates space for a resource from the specified bus address region, and changes pci_bus_alloc_resource() to allocate space for 64-bit BARs from the entire bus address region, and space for 32-bit BARs from only the bus address region below 4GB. If we had this window: pci_root HWP0002:0a: host bridge window [mem 0xf0180000000-0xf01fedfffff] (bus address [0x80000000-0xfedfffff]) we previously could not put a 32-bit BAR there, because the CPU addresses don't fit in 32 bits. This patch fixes this, so we can use this space for 32-bit BARs. It's also possible (though unlikely) to have resources with 32-bit CPU addresses but bus addresses above 4GB. In this case the previous code would allocate space that a 32-bit BAR could not map. Remove PCIBIOS_MAX_MEM_32, which is no longer used. [bhelgaas: reworked starting from http://lkml.kernel.org/r/1386658484-15774-3-git-send-email-yinghai@kernel.org] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-12-20 16:57:37 +00:00
#endif
return pci_bus_alloc_from_region(bus, res, size, align, min,
type_mask, alignf, alignf_data,
&pci_32_bit);
}
EXPORT_SYMBOL(pci_bus_alloc_resource);
/*
* The @idx resource of @dev should be a PCI-PCI bridge window. If this
* resource fits inside a window of an upstream bridge, do nothing. If it
* overlaps an upstream window but extends outside it, clip the resource so
* it fits completely inside.
*/
bool pci_bus_clip_resource(struct pci_dev *dev, int idx)
{
struct pci_bus *bus = dev->bus;
struct resource *res = &dev->resource[idx];
struct resource orig_res = *res;
struct resource *r;
pci_bus_for_each_resource(bus, r) {
resource_size_t start, end;
if (!r)
continue;
if (resource_type(res) != resource_type(r))
continue;
start = max(r->start, res->start);
end = min(r->end, res->end);
if (start > end)
continue; /* no overlap */
if (res->start == start && res->end == end)
return false; /* no change */
res->start = start;
res->end = end;
PCI: Clear IORESOURCE_UNSET when clipping a bridge window c770cb4cb505 ("PCI: Mark invalid BARs as unassigned") sets IORESOURCE_UNSET if we fail to claim a resource. If we tried to claim a bridge window, failed, clipped the window, and tried to claim the clipped window, we failed again because of IORESOURCE_UNSET: pci_bus 0000:00: root bus resource [mem 0xc0000000-0xffffffff window] pci 0000:00:01.0: can't claim BAR 15 [mem 0xbdf00000-0xddefffff 64bit pref]: no compatible bridge window pci 0000:00:01.0: [mem size 0x20000000 64bit pref] clipped to [mem size 0x1df00000 64bit pref] pci 0000:00:01.0: bridge window [mem size 0x1df00000 64bit pref] pci 0000:00:01.0: can't claim BAR 15 [mem size 0x1df00000 64bit pref]: no address assigned The 00:01.0 window started as [mem 0xbdf00000-0xddefffff 64bit pref]. That starts before the host bridge window [mem 0xc0000000-0xffffffff window], so we clipped the 00:01.0 window to [mem 0xc0000000-0xddefffff 64bit pref]. But we left it marked IORESOURCE_UNSET, so the second claim failed when it should have succeeded. This means downstream devices will also fail for lack of resources, e.g., in the bugzilla below, radeon 0000:01:00.0: Fatal error during GPU init Clear IORESOURCE_UNSET when we clip a bridge window. Also clear IORESOURCE_UNSET in our copy of the unclipped window so we can see exactly what the original window was and how it now fits inside the upstream window. Fixes: c770cb4cb505 ("PCI: Mark invalid BARs as unassigned") Link: https://bugzilla.kernel.org/show_bug.cgi?id=85491#c47 Based-on-patch-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Based-on-patch-by: Yinghai Lu <yinghai@kernel.org> Tested-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Acked-by: Yinghai Lu <yinghai@kernel.org> CC: stable@vger.kernel.org # v4.1+
2015-09-22 22:03:54 +00:00
res->flags &= ~IORESOURCE_UNSET;
orig_res.flags &= ~IORESOURCE_UNSET;
pci_info(dev, "%pR clipped to %pR\n", &orig_res, res);
return true;
}
return false;
}
void __weak pcibios_resource_survey_bus(struct pci_bus *bus) { }
void __weak pcibios_bus_add_device(struct pci_dev *pdev) { }
/**
* pci_bus_add_device - start driver for a single device
* @dev: device to add
*
* This adds add sysfs entries and start device drivers
*/
void pci_bus_add_device(struct pci_dev *dev)
{
PCI: move OF status = "disabled" detection to dev->match_driver The blamed commit has broken probing on arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi when &enetc_port0 (PCI function 0) has status = "disabled". Background: pci_scan_slot() has logic to say that if the function 0 of a device is absent, the entire device is absent and we can skip the other functions entirely. Traditionally, this has meant that pci_bus_read_dev_vendor_id() returns an error code for that function. However, since the blamed commit, there is an extra confounding condition: function 0 of the device exists and has a valid vendor id, but it is disabled in the device tree. In that case, pci_scan_slot() would incorrectly skip the entire device instead of just that function. In the case of NXP LS1028A, status = "disabled" does not mean that the PCI function's config space is not available for reading. It is, but the Ethernet port is just not functionally useful with a particular SerDes protocol configuration (0x9999) due to pinmuxing constraints of the Soc. So, pci_scan_slot() skips all other functions on the ENETC ECAM (enetc_port1, enetc_port2, enetc_mdio_pf3 etc) when just enetc_port0 had to not be probed. There is an additional regression introduced by the change, caused by its fundamental premise. The enetc driver needs to run code for all PCI functions, regardless of whether they're enabled or not in the device tree. That is no longer possible if the driver's probe function is no longer called. But Rob recommends that we move the of_device_is_available() detection to dev->match_driver, and this makes the PCI fixups still run on all functions, while just probing drivers for those functions that are enabled. So, a separate change in the enetc driver will have to move the workarounds to a PCI fixup. Fixes: 6fffbc7ae137 ("PCI: Honor firmware's device disabled status") Link: https://lore.kernel.org/netdev/CAL_JsqLsVYiPLx2kcHkDQ4t=hQVCR7NHziDwi9cCFUFhx48Qow@mail.gmail.com/ Suggested-by: Rob Herring <robh@kernel.org> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2023-08-03 13:58:56 +00:00
struct device_node *dn = dev->dev.of_node;
int retval;
/*
* Can not put in pci_device_add yet because resources
* are not assigned yet for some devices.
*/
pcibios_bus_add_device(dev);
pci_fixup_device(pci_fixup_final, dev);
2023-08-15 17:19:57 +00:00
if (pci_is_bridge(dev))
of_pci_make_dev_node(dev);
pci_create_sysfs_dev_files(dev);
pci_proc_attach_device(dev);
pci_bridge_d3_update(dev);
PCI: move OF status = "disabled" detection to dev->match_driver The blamed commit has broken probing on arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi when &enetc_port0 (PCI function 0) has status = "disabled". Background: pci_scan_slot() has logic to say that if the function 0 of a device is absent, the entire device is absent and we can skip the other functions entirely. Traditionally, this has meant that pci_bus_read_dev_vendor_id() returns an error code for that function. However, since the blamed commit, there is an extra confounding condition: function 0 of the device exists and has a valid vendor id, but it is disabled in the device tree. In that case, pci_scan_slot() would incorrectly skip the entire device instead of just that function. In the case of NXP LS1028A, status = "disabled" does not mean that the PCI function's config space is not available for reading. It is, but the Ethernet port is just not functionally useful with a particular SerDes protocol configuration (0x9999) due to pinmuxing constraints of the Soc. So, pci_scan_slot() skips all other functions on the ENETC ECAM (enetc_port1, enetc_port2, enetc_mdio_pf3 etc) when just enetc_port0 had to not be probed. There is an additional regression introduced by the change, caused by its fundamental premise. The enetc driver needs to run code for all PCI functions, regardless of whether they're enabled or not in the device tree. That is no longer possible if the driver's probe function is no longer called. But Rob recommends that we move the of_device_is_available() detection to dev->match_driver, and this makes the PCI fixups still run on all functions, while just probing drivers for those functions that are enabled. So, a separate change in the enetc driver will have to move the workarounds to a PCI fixup. Fixes: 6fffbc7ae137 ("PCI: Honor firmware's device disabled status") Link: https://lore.kernel.org/netdev/CAL_JsqLsVYiPLx2kcHkDQ4t=hQVCR7NHziDwi9cCFUFhx48Qow@mail.gmail.com/ Suggested-by: Rob Herring <robh@kernel.org> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2023-08-03 13:58:56 +00:00
dev->match_driver = !dn || of_device_is_available(dn);
retval = device_attach(&dev->dev);
if (retval < 0 && retval != -EPROBE_DEFER)
pci_warn(dev, "device attach failed (%d)\n", retval);
pci_dev_assign_added(dev, true);
if (dev_of_node(&dev->dev) && pci_is_bridge(dev)) {
retval = of_platform_populate(dev_of_node(&dev->dev), NULL, NULL,
&dev->dev);
if (retval)
pci_err(dev, "failed to populate child OF nodes (%d)\n",
retval);
}
}
EXPORT_SYMBOL_GPL(pci_bus_add_device);
/**
* pci_bus_add_devices - start driver for PCI devices
* @bus: bus to check for new devices
*
* Start driver for PCI devices and add some sysfs entries.
*/
void pci_bus_add_devices(const struct pci_bus *bus)
{
struct pci_dev *dev;
struct pci_bus *child;
list_for_each_entry(dev, &bus->devices, bus_list) {
/* Skip already-added devices */
if (pci_dev_is_added(dev))
continue;
pci_bus_add_device(dev);
}
list_for_each_entry(dev, &bus->devices, bus_list) {
/* Skip if device attach failed */
if (!pci_dev_is_added(dev))
continue;
child = dev->subordinate;
if (child)
pci_bus_add_devices(child);
}
}
EXPORT_SYMBOL(pci_bus_add_devices);
PCI/ASPM: Fix deadlock when enabling ASPM A last minute revert in 6.7-final introduced a potential deadlock when enabling ASPM during probe of Qualcomm PCIe controllers as reported by lockdep: ============================================ WARNING: possible recursive locking detected 6.7.0 #40 Not tainted -------------------------------------------- kworker/u16:5/90 is trying to acquire lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pcie_aspm_pm_state_change+0x58/0xdc but task is already holding lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pci_walk_bus+0x34/0xbc other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(pci_bus_sem); lock(pci_bus_sem); *** DEADLOCK *** Call trace: print_deadlock_bug+0x25c/0x348 __lock_acquire+0x10a4/0x2064 lock_acquire+0x1e8/0x318 down_read+0x60/0x184 pcie_aspm_pm_state_change+0x58/0xdc pci_set_full_power_state+0xa8/0x114 pci_set_power_state+0xc4/0x120 qcom_pcie_enable_aspm+0x1c/0x3c [pcie_qcom] pci_walk_bus+0x64/0xbc qcom_pcie_host_post_init_2_7_0+0x28/0x34 [pcie_qcom] The deadlock can easily be reproduced on machines like the Lenovo ThinkPad X13s by adding a delay to increase the race window during asynchronous probe where another thread can take a write lock. Add a new pci_set_power_state_locked() and associated helper functions that can be called with the PCI bus semaphore held to avoid taking the read lock twice. Link: https://lore.kernel.org/r/ZZu0qx2cmn7IwTyQ@hovoldconsulting.com Link: https://lore.kernel.org/r/20240130100243.11011-1-johan+linaro@kernel.org Fixes: f93e71aea6c6 ("Revert "PCI/ASPM: Remove pcie_aspm_pm_state_change()"") Signed-off-by: Johan Hovold <johan+linaro@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Cc: <stable@vger.kernel.org> # 6.7
2024-01-30 10:02:43 +00:00
static void __pci_walk_bus(struct pci_bus *top, int (*cb)(struct pci_dev *, void *),
void *userdata, bool locked)
{
struct pci_dev *dev;
struct pci_bus *bus;
struct list_head *next;
int retval;
bus = top;
PCI/ASPM: Fix deadlock when enabling ASPM A last minute revert in 6.7-final introduced a potential deadlock when enabling ASPM during probe of Qualcomm PCIe controllers as reported by lockdep: ============================================ WARNING: possible recursive locking detected 6.7.0 #40 Not tainted -------------------------------------------- kworker/u16:5/90 is trying to acquire lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pcie_aspm_pm_state_change+0x58/0xdc but task is already holding lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pci_walk_bus+0x34/0xbc other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(pci_bus_sem); lock(pci_bus_sem); *** DEADLOCK *** Call trace: print_deadlock_bug+0x25c/0x348 __lock_acquire+0x10a4/0x2064 lock_acquire+0x1e8/0x318 down_read+0x60/0x184 pcie_aspm_pm_state_change+0x58/0xdc pci_set_full_power_state+0xa8/0x114 pci_set_power_state+0xc4/0x120 qcom_pcie_enable_aspm+0x1c/0x3c [pcie_qcom] pci_walk_bus+0x64/0xbc qcom_pcie_host_post_init_2_7_0+0x28/0x34 [pcie_qcom] The deadlock can easily be reproduced on machines like the Lenovo ThinkPad X13s by adding a delay to increase the race window during asynchronous probe where another thread can take a write lock. Add a new pci_set_power_state_locked() and associated helper functions that can be called with the PCI bus semaphore held to avoid taking the read lock twice. Link: https://lore.kernel.org/r/ZZu0qx2cmn7IwTyQ@hovoldconsulting.com Link: https://lore.kernel.org/r/20240130100243.11011-1-johan+linaro@kernel.org Fixes: f93e71aea6c6 ("Revert "PCI/ASPM: Remove pcie_aspm_pm_state_change()"") Signed-off-by: Johan Hovold <johan+linaro@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Cc: <stable@vger.kernel.org> # 6.7
2024-01-30 10:02:43 +00:00
if (!locked)
down_read(&pci_bus_sem);
next = top->devices.next;
for (;;) {
if (next == &bus->devices) {
/* end of this bus, go up or finish */
if (bus == top)
break;
next = bus->self->bus_list.next;
bus = bus->self->bus;
continue;
}
dev = list_entry(next, struct pci_dev, bus_list);
if (dev->subordinate) {
/* this is a pci-pci bridge, do its devices next */
next = dev->subordinate->devices.next;
bus = dev->subordinate;
} else
next = dev->bus_list.next;
retval = cb(dev, userdata);
if (retval)
break;
}
PCI/ASPM: Fix deadlock when enabling ASPM A last minute revert in 6.7-final introduced a potential deadlock when enabling ASPM during probe of Qualcomm PCIe controllers as reported by lockdep: ============================================ WARNING: possible recursive locking detected 6.7.0 #40 Not tainted -------------------------------------------- kworker/u16:5/90 is trying to acquire lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pcie_aspm_pm_state_change+0x58/0xdc but task is already holding lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pci_walk_bus+0x34/0xbc other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(pci_bus_sem); lock(pci_bus_sem); *** DEADLOCK *** Call trace: print_deadlock_bug+0x25c/0x348 __lock_acquire+0x10a4/0x2064 lock_acquire+0x1e8/0x318 down_read+0x60/0x184 pcie_aspm_pm_state_change+0x58/0xdc pci_set_full_power_state+0xa8/0x114 pci_set_power_state+0xc4/0x120 qcom_pcie_enable_aspm+0x1c/0x3c [pcie_qcom] pci_walk_bus+0x64/0xbc qcom_pcie_host_post_init_2_7_0+0x28/0x34 [pcie_qcom] The deadlock can easily be reproduced on machines like the Lenovo ThinkPad X13s by adding a delay to increase the race window during asynchronous probe where another thread can take a write lock. Add a new pci_set_power_state_locked() and associated helper functions that can be called with the PCI bus semaphore held to avoid taking the read lock twice. Link: https://lore.kernel.org/r/ZZu0qx2cmn7IwTyQ@hovoldconsulting.com Link: https://lore.kernel.org/r/20240130100243.11011-1-johan+linaro@kernel.org Fixes: f93e71aea6c6 ("Revert "PCI/ASPM: Remove pcie_aspm_pm_state_change()"") Signed-off-by: Johan Hovold <johan+linaro@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Cc: <stable@vger.kernel.org> # 6.7
2024-01-30 10:02:43 +00:00
if (!locked)
up_read(&pci_bus_sem);
}
/**
* pci_walk_bus - walk devices on/under bus, calling callback.
* @top: bus whose devices should be walked
* @cb: callback to be called for each device found
* @userdata: arbitrary pointer to be passed to callback
*
* Walk the given bus, including any bridged devices
* on buses under this bus. Call the provided callback
* on each device found.
*
* We check the return of @cb each time. If it returns anything
* other than 0, we break out.
*/
void pci_walk_bus(struct pci_bus *top, int (*cb)(struct pci_dev *, void *), void *userdata)
{
__pci_walk_bus(top, cb, userdata, false);
}
EXPORT_SYMBOL_GPL(pci_walk_bus);
PCI/ASPM: Fix deadlock when enabling ASPM A last minute revert in 6.7-final introduced a potential deadlock when enabling ASPM during probe of Qualcomm PCIe controllers as reported by lockdep: ============================================ WARNING: possible recursive locking detected 6.7.0 #40 Not tainted -------------------------------------------- kworker/u16:5/90 is trying to acquire lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pcie_aspm_pm_state_change+0x58/0xdc but task is already holding lock: ffffacfa78ced000 (pci_bus_sem){++++}-{3:3}, at: pci_walk_bus+0x34/0xbc other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(pci_bus_sem); lock(pci_bus_sem); *** DEADLOCK *** Call trace: print_deadlock_bug+0x25c/0x348 __lock_acquire+0x10a4/0x2064 lock_acquire+0x1e8/0x318 down_read+0x60/0x184 pcie_aspm_pm_state_change+0x58/0xdc pci_set_full_power_state+0xa8/0x114 pci_set_power_state+0xc4/0x120 qcom_pcie_enable_aspm+0x1c/0x3c [pcie_qcom] pci_walk_bus+0x64/0xbc qcom_pcie_host_post_init_2_7_0+0x28/0x34 [pcie_qcom] The deadlock can easily be reproduced on machines like the Lenovo ThinkPad X13s by adding a delay to increase the race window during asynchronous probe where another thread can take a write lock. Add a new pci_set_power_state_locked() and associated helper functions that can be called with the PCI bus semaphore held to avoid taking the read lock twice. Link: https://lore.kernel.org/r/ZZu0qx2cmn7IwTyQ@hovoldconsulting.com Link: https://lore.kernel.org/r/20240130100243.11011-1-johan+linaro@kernel.org Fixes: f93e71aea6c6 ("Revert "PCI/ASPM: Remove pcie_aspm_pm_state_change()"") Signed-off-by: Johan Hovold <johan+linaro@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Cc: <stable@vger.kernel.org> # 6.7
2024-01-30 10:02:43 +00:00
void pci_walk_bus_locked(struct pci_bus *top, int (*cb)(struct pci_dev *, void *), void *userdata)
{
lockdep_assert_held(&pci_bus_sem);
__pci_walk_bus(top, cb, userdata, true);
}
EXPORT_SYMBOL_GPL(pci_walk_bus_locked);
struct pci_bus *pci_bus_get(struct pci_bus *bus)
{
if (bus)
get_device(&bus->dev);
return bus;
}
void pci_bus_put(struct pci_bus *bus)
{
if (bus)
put_device(&bus->dev);
}