atmel-ssc-dai problem with ssc1
Raphaël Poggi
poggi.raph at gmail.com
Sat Apr 16 04:32:42 PDT 2016
Hi,
If I remember correctly, this problem is old, but there is a patch
([1]) which fix this. Did you try it ?
[1]: http://lkml.iu.edu/hypermail/linux/kernel/1601.1/04004.html
2016-04-15 23:47 GMT+02:00 Peter Rosin <peda at axentia.se>:
> Hi!
>
> I have problems using the sound/soc/atmel/atmel_ssc_dai.c driver with
> ssc1 on a board derived from the at91sam9261ek board. We use a different
> audio codec and we use ssc1 instead of ssc0, but that really should not
> matter. But it does.
>
> I have copied shamelessly from the sound/soc/atmel/atmel_wm8904.c card
> driver, with one of the changes showing a big problem in the atmel_ssc_dai
> driver. I call atmel_ssc_set_audio(1) where atmel_wm8904.c calls
> atmel_ssc_set_audio(0). Well, both call a value given from dt, but the
> dts using the atmel_wm8904 driver all use ssc0 and my dts uses ssc1.
>
> It turns out that all the dai devices registered by atmel_ssc_dai end
> up with id 0, which does not work at all in e.g.
> atmel_ssc_dai.c:atmel_ssc_startup() which is using dai->id as index
> into the ssc_info[] array and therefore looks at index 0 when the setup
> by atmel_ssc_set_audio(1) had stored the needed stuff at index 1.
>
> This leads to a NULL dereference, such as:
>
> Unable to handle kernel NULL pointer dereference at virtual address 0000000c
> pgd = c1bf8000
> [0000000c] *pgd=218e9831, *pte=00000000, *ppte=00000000
> Internal error: Oops: 17 [#1] ARM
> Modules linked in:
> CPU: 0 PID: 517 Comm: aplay Not tainted 4.1.0+ #20
> Hardware name: Atmel AT91SAM9
> task: c19e3180 ti: c1bee000 task.ti: c1bee000
> PC is at atmel_ssc_startup+0x1c/0x14c
> LR is at soc_pcm_open+0xd4/0x7c8
> pc : [<c030155c>] lr : [<c02f8e0c>] psr: a0000013
> sp : c1befcd8 ip : c1a08ab8 fp : 00000000
> r10: c1a0fc00 r9 : c1beff74 r8 : 00000000
> r7 : c0597d50 r6 : c19b5c00 r5 : c19b5c00 r4 : c0597d50
> r3 : 00000000 r2 : 00000000 r1 : c1a065e0 r0 : c19b5c00
> Flags: NzCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user
> Control: 0005317f Table: 21bf8000 DAC: 00000015
> Process aplay (pid: 517, stack limit = 0xc1bee190)
> Stack: (0xc1befcd8 to 0xc1bf0000)
> fcc0: c19b5c00 c1a065e0
> fce0: 00000000 c0301540 c19b6810 c19b5c00 c1a065e0 00000001 00000000 c1beff74
> fd00: c1a0fc00 c02f8e0c c1a0fc00 00000000 ffffffff 00000011 c19ab400 000f4240
> fd20: 00000009 c02e4d90 000f4240 c19b681c 0000000b ffffffff 00000000 c1befd8c
> fd40: 00000000 c19b7f88 c19b7f94 00000000 c1beff74 00000000 00000000 c02e4e00
> fd60: c19b7c00 c19b5c00 c1a375c0 c19b7c00 c1a375c0 c02e4ee4 00000000 c19e3180
> fd80: c002f938 c19b7f94 c19b7f94 c19b7c00 00000010 c19b7c00 c1a375c0 00000000
> fda0: c1a22ce0 07400010 c1977000 c02e50b0 c0408f30 c1a375c0 c198b8b8 c02d8e54
> fdc0: c198b8b8 c1a375c0 c191e680 00000000 c0408a40 c0087784 c198b8b8 00000010
> fde0: c1a375c0 c198b8b8 c008767c c1a375c8 c1a375c0 c0082f6c c1befef0 c1a375c0
> fe00: c1befeb4 c1befebc c1a375c0 c008fe94 00000037 c1be34a8 b6f10000 ffffffff
> fe20: 00000000 00000000 00020802 c1be34a8 c198bc60 00000000 c1414c38 00000000
> fe40: 00000041 00000026 00000000 00000000 c1977000 00500088 c1977000 c008df24
> fe60: 00000000 00500088 c1befe98 c00809e0 c198b8b8 00000002 00000040 c1befef0
> fe80: c1a375c0 c1beff74 ffffff9c 00000000 c1bee000 c1977000 00000041 c0090270
> fea0: c1befebc c1977000 c1a32d00 c02dd728 c1b527d4 c1ad4030 c1417198 00000000
> fec0: 00000054 c1beff74 00000001 ffffff9c c1befef0 c1977000 c1bee000 00000000
> fee0: 00000000 c00905e4 00000041 00000005 c1ad4030 c1417198 c4254b62 00000008
> ff00: c1977019 be82833c 00000000 c1440bb0 c198b8b8 00000101 00000004 00000040
> ff20: 00000000 00000000 00000000 00000000 c1b39020 00000000 00000400 c1bc5660
> ff40: 000a0802 c009ad10 000a0802 ffffff9c 00000004 ffffff9c c1977000 00000005
> ff60: c000a384 c0082b94 00000000 c1a37660 c1a37660 00020802 40040000 00000026
> ff80: 00000100 00000001 0002c6f8 be82848c 00000000 ffffffff 00000005 c000a384
> ffa0: 00000000 c000a200 be82848c 00000000 be82848c 000a0802 00000000 00000000
> ffc0: be82848c 00000000 ffffffff 00000005 00000000 00000000 00000001 00000000
> ffe0: b6f9cee4 be828328 b6f33284 b6e1bccc 60000010 be82848c 21ffd831 21ffdc31
> [<c030155c>] (atmel_ssc_startup) from [<c02f8e0c>] (soc_pcm_open+0xd4/0x7c8)
> [<c02f8e0c>] (soc_pcm_open) from [<c02e4e00>] (snd_pcm_open_substream+0x48/0x90)
> [<c02e4e00>] (snd_pcm_open_substream) from [<c02e4ee4>] (snd_pcm_open+0x9c/0x1cc)
> [<c02e4ee4>] (snd_pcm_open) from [<c02e50b0>] (snd_pcm_playback_open+0x40/0x5c)
> [<c02e50b0>] (snd_pcm_playback_open) from [<c02d8e54>] (snd_open+0x128/0x148)
> [<c02d8e54>] (snd_open) from [<c0087784>] (chrdev_open+0x108/0x12c)
> [<c0087784>] (chrdev_open) from [<c0082f6c>] (do_dentry_open+0x1bc/0x2d0)
> [<c0082f6c>] (do_dentry_open) from [<c008fe94>] (do_last+0x930/0xb58)
> [<c008fe94>] (do_last) from [<c0090270>] (path_openat+0x1b4/0x4f8)
> [<c0090270>] (path_openat) from [<c00905e4>] (do_filp_open+0x30/0x7c)
> [<c00905e4>] (do_filp_open) from [<c0082b94>] (do_sys_open+0x124/0x1c4)
> [<c0082b94>] (do_sys_open) from [<c000a200>] (ret_fast_syscall+0x0/0x38)
> Code: e3a0403c e0247493 e5943004 e1a06000 (e593300c)
> ---[ end trace edfb18e35da5cede ]---
>
> I'm fairly confident that this comes from the line
> clk_enable(ssc_p->ssc->clk);
> in atmel_ssc_startup() when ssc_p->ssc is NULL.
>
> Looking further, the function that selects id 0 is fmt_single_name() in
> sound/soc/soc-core.c which would have found individual ids for the three
> ssc devices if they would have had names like "fffbc000.ssc.0",
> "fffc0000.ssc.1" and "fffc4000.ssc.2" instead of plain "fffbc000.ssc" etc
> (i.e. w/o the needed trailing .#) which is what they do have as dev_name.
>
> I can force sound to work on my board by installing the below revolting
> patch (there is only one dai in use with legacy dai naming) which is
> a pretty good indicator that my analysis is correct.
>
> The atmel_ssc_dai driver really should make sure that the dais it registers
> end up with the correct id. Or, it could be the case that at91sam9261.dtsi
> is at fault for giving the ssc:s bad names, but something is clearly not
> working as intended.
>
> BTW, I'm on d034050b9d4d3d73 on the linux-4.1-at91 branch (nov 2015) of
> the linux4sam/linux-at91.git tree at github, plus a bunch of local patches.
>
> Cheers,
> Peter
>
> diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c
> index 2373252..671a63e 100644 (file)
> --- a/sound/soc/soc-core.c
> +++ b/sound/soc/soc-core.c
> @@ -2601,6 +2601,8 @@ static int snd_soc_register_dais(struct snd_soc_component *component,
> */
> if (count == 1 && legacy_dai_naming) {
> dai->name = fmt_single_name(dev, &dai->id);
> + /* uurk, force id 1 since we use ssc1 */
> + dai->id = 1;
> } else {
> dai->name = fmt_multiple_name(dev, &dai_drv[i]);
> if (dai_drv[i].id)
>
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
More information about the linux-arm-kernel
mailing list