ALSA: emu10k1: fix writing 1st pointer-offset register set through /proc

The limits were appropriate only for the 2nd set.

FWIW, the channel count 4 for the 2nd set is suspicious as well - at
least P17V_PLAYBACK_FIFO_PTR actually has 8 channels, and comments on
HCFG2 hint at that as well. But all bitmasks are documented only for 4
channels. Anyway, rectifying that is out of scope for this patch.

Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230526101659.437969-3-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This commit is contained in:
Oswald Buddenhagen 2023-05-26 12:16:55 +02:00 committed by Takashi Iwai
parent 219153c6ed
commit 67ff2add9e

View File

@ -483,7 +483,8 @@ static void snd_emu_proc_ptr_reg_read(struct snd_info_entry *entry,
}
static void snd_emu_proc_ptr_reg_write(struct snd_info_entry *entry,
struct snd_info_buffer *buffer, int iobase)
struct snd_info_buffer *buffer,
int iobase, int length, int voices)
{
struct snd_emu10k1 *emu = entry->private_data;
char line[64];
@ -491,7 +492,7 @@ static void snd_emu_proc_ptr_reg_write(struct snd_info_entry *entry,
while (!snd_info_get_line(buffer, line, sizeof(line))) {
if (sscanf(line, "%x %x %x", &reg, &channel_id, &val) != 3)
continue;
if (reg < 0xa0 && val <= 0xffffffff && channel_id <= 3)
if (reg < length && channel_id < voices)
snd_ptr_write(emu, iobase, reg, channel_id, val);
}
}
@ -499,13 +500,15 @@ static void snd_emu_proc_ptr_reg_write(struct snd_info_entry *entry,
static void snd_emu_proc_ptr_reg_write00(struct snd_info_entry *entry,
struct snd_info_buffer *buffer)
{
snd_emu_proc_ptr_reg_write(entry, buffer, 0);
snd_emu_proc_ptr_reg_write(entry, buffer, 0, 0x80, 64);
}
static void snd_emu_proc_ptr_reg_write20(struct snd_info_entry *entry,
struct snd_info_buffer *buffer)
{
snd_emu_proc_ptr_reg_write(entry, buffer, 0x20);
struct snd_emu10k1 *emu = entry->private_data;
snd_emu_proc_ptr_reg_write(entry, buffer, 0x20,
emu->card_capabilities->ca0108_chip ? 0xa0 : 0x80, 4);
}