GitLab

After hours of debugging, I finally found the reason why some source
and runtime combination does not work. The PTP (page table pages)
address must be aligned. I am not sure how much, but alignment to
PAGE_SIZE is sufficient. Also, use ALSA's page allocation routines
to ensure proper virtual -> physical address translation.

Cc: <stable@kernel.org>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>

The variables are unsigned so the test `>= 0' is always true,
the `< 0' test always fails. In these cases the other part of
the test catches wrapped values.

In dac_audio_write() there does not occur a test for wrapped
values, but the test appears redundant.
Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

On Soundblaster X-FI Titenium with emu20k2 the SIDE and SURROUND mute
functions are swapped.
It was checked with 'speaker-test -c 8 -s 3' and (un)mute surround or
'speaker-test -c 8 -s 7' and (un)mute side. The volume seems not
to be affected and works as expected.
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

- replace NULL == xxx with !xxx
- replace NULL != xxx with xxx
- similar trivial cleanups
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Added the suspend/resume support to ctxfi driver.

The team tested on the following seems ok:
  AMD Athlon 64 3500+ / ASUS A8N-E / 512MB DDR ATI / Radeon X1300
  20k1 & 20k2 cards
Signed-off-by: Wai Yew CHAY <wychay@ctl.creative.com>
Singed-off-by: Ryan RICHARDS <ryan_richards@creativelabs.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Allow unknown PCI SSIDs for emu20k1 and emu20k2 as "unknown" model.
Also, add a black-list check in case any device has to be listed
as "unsupported".  It has a negative value in the pci quirk entry.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The spinlock in atc can cause a sleep in lock:

Kernel failure message 1:
BUG: sleeping function called from invalid context at mm/slub.c:1599
in_atomic(): 0, irqs_disabled(): 1, pid: 2537, name: gstreamer-prope
Pid: 2537, comm: gstreamer-prope Tainted: P
2.6.29.4-167.fc11.x86_64 #1
Call Trace:
  [<ffffffff8103ff0f>] __might_sleep+0x10b/0x110
  [<ffffffff810cd734>] __kmalloc+0x73/0x130
  [<ffffffffa0b4b142>] ? daio_rsc_init+0xaa/0x125 [snd_ctxfi]
  [<ffffffffa0b4b212>] dao_rsc_init+0x55/0x1c0 [snd_ctxfi]
  [<ffffffffa0b4b3d2>] dao_rsc_reinit+0x55/0x5d [snd_ctxfi]
  [<ffffffff813abd6c>] ? _spin_lock_irqsave+0x32/0x3b
  [<ffffffffa0b454fe>] atc_spdif_out_passthru+0x92/0x136 [snd_ctxfi]
  ...

Since the lock path is no critical path, it can be gracefully
replaced with a mutex.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Currently the PCM resources are allocated only once and ever in prepare
callback, assuming that the PCM parameters are never changed.  But it's
not true.

This patch adds the call of atc->pcm_release_resources() at hw_params
and hw_free callbacks to assure that the PCM setup is done correctly
for each h/w parameter changes.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The SRC instances may not exist when PCM pointer callback is called at
the state before initialization is finished.  Add the NULL check just
to be sure.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Takashi Iwai <tiwai@suse.de>

CTUAA should be checked instead of CTHENDRIX.  The latter is for 20k2 chip.
Also, fixed the detection of UAA/HENDRIX models by fixing the mask bits.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Clean up probe routines and model detection routines so that the driver
won't call and check the PCI subsystem id at each time.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Remove useless variable initializations and cast at the beginning of
functions.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

- Use static tables instead of assigining each funciton pointer
- Add __devinit* to appropriate places; pcm, mixer and timer cannot be
  marked because they are kept in the function table that lives long
- Move create_alsa_devs function out of struct ct_atc to mark it
  __devinit
Signed-off-by: Takashi Iwai <tiwai@suse.de>

emu20k1 has a native timer interrupt based on the audio clock, which
is more accurate than the system timer (from the synchronization POV).
This patch adds the code to handle this with multiple streams.

The system timer is still used on emu20k2, and can be used also for
emu20k1 easily by changing USE_SYSTEM_TIMER to 1 in cttimer.c.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Use SG-buffers instead of contiguous pages.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The device seems supporting only U8, S16, S24_3LE, S32.  Other linear
formats result in bad outputs.

Also, added the support for 32bit float format, which wasn't listed
in the original code.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

PCM names for surround streams should be also fixed as well as the mixer
element names.  Also, a bit clean up for PCM name setup.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The prepare callback can be called multiple times, thus it needs to
release and acquire the resource again by itself at the second or later
call.

Simply add pcm_release_resources() at the beginning of each prepare
callback in ctatc.c.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Replace a spinlock with a mutex protecting the vm block list at
mmap / munmap calls, which caused Oops like below:

BUG: sleeping function called from invalid context at mm/slub.c:1599
in_atomic(): 0, irqs_disabled(): 1, pid: 32065, name: xine
Pid: 32065, comm: xine Tainted: P           2.6.29.4-75.fc10.x86_64 #1
Call Trace:
  [<ffffffff81040685>] __might_sleep+0x105/0x10a
  [<ffffffff810c9fae>] kmem_cache_alloc+0x32/0xe2
  [<ffffffffa08e3110>] ct_vm_map+0xfa/0x19e [snd_ctxfi]
  [<ffffffffa08e1a07>] ct_map_audio_buffer+0x4c/0x76 [snd_ctxfi]
  [<ffffffffa08e2aa5>] atc_pcm_playback_prepare+0x1d7/0x2a8 [snd_ctxfi]
  [<ffffffff8105ef3f>] ? up_read+0x9/0xb
  [<ffffffff81186b61>] ? __up_read+0x7c/0x87
  [<ffffffffa08e36a6>] ct_pcm_playback_prepare+0x39/0x60 [snd_ctxfi]
  [<ffffffffa0886bcb>] snd_pcm_do_prepare+0x16/0x28 [snd_pcm]
  [<ffffffffa08867c7>] snd_pcm_action_single+0x2d/0x5b [snd_pcm]
  [<ffffffffa08881f3>] snd_pcm_action_nonatomic+0x52/0x6a [snd_pcm]
  [<ffffffffa088a723>] snd_pcm_common_ioctl1+0x404/0xc79 [snd_pcm]
  [<ffffffff810c52c8>] ? alloc_pages_current+0xb9/0xc2
  [<ffffffff810c9402>] ? new_slab+0x1a5/0x1cb
  [<ffffffff810ab9ea>] ? vma_prio_tree_insert+0x23/0xc1
  [<ffffffffa088b411>] snd_pcm_playback_ioctl1+0x213/0x230 [snd_pcm]
  [<ffffffff810b6c20>] ? mmap_region+0x397/0x4c9
  [<ffffffffa088bd9b>] snd_pcm_playback_ioctl+0x2e/0x36 [snd_pcm]
  [<ffffffff810ddc64>] vfs_ioctl+0x2a/0x78
  [<ffffffff810de130>] do_vfs_ioctl+0x462/0x4a2
  [<ffffffff81029cef>] ? default_spin_lock_flags+0x9/0xe
  [<ffffffff81374647>] ? trace_hardirqs_off_thunk+0x3a/0x6c
  [<ffffffff810de1c5>] sys_ioctl+0x55/0x77
  [<ffffffff8101133a>] system_call_fastpath+0x16/0x1b
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Added ctxfi: prefix to each debug print.
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Use struct pci subsystem_device and revision fields instead of
unneeded calls of pci_read_config_*().
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Takashi Iwai <tiwai@suse.de>

The Sound Blaster X-Fi driver supports Creative solutions based on
20K1 and 20K2 chipsets.

Supported hardware :

Creative Sound Blaster X-Fi Titanium Fatal1ty® Champion Series
Creative Sound Blaster X-Fi Titanium Fatal1ty Professional Series
Creative Sound Blaster X-Fi Titanium Professional Audio
Creative Sound Blaster X-Fi Titanium
Creative Sound Blaster X-Fi Elite Pro
Creative Sound Blaster X-Fi Platinum
Creative Sound Blaster X-Fi Fatal1ty
Creative Sound Blaster X-Fi XtremeGamer
Creative Sound Blaster X-Fi XtremeMusic

Current release features:

* ALSA PCM Playback
* ALSA Record
* ALSA Mixer

Note:

* External I/O modules detection not included.
Signed-off-by: Wai Yew CHAY <wychay@ctl.creative.com>
Singed-off-by: Ryan RICHARDS <ryan_richards@creativelabs.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>