[alsa-devel] [PATCH] ALSA: hda/tegra - Fix oops

[Thierry Reding]

From: Thierry Reding <treding at nvidia.com>

Commit a41d122449be ("ALSA: hda - Embed bus into controller object")
introduced a regression in the Tegra HDA driver that causes the
following oops during boot:

	[    2.333458] Unable to handle kernel NULL pointer dereference at virtual address 000004c4
	[    2.341537] pgd = c0004000
	[    2.344312] [000004c4] *pgd=00000000
	[    2.347898] Internal error: Oops: 5 [#1] PREEMPT SMP ARM
	[    2.353200] Modules linked in:
	[    2.356264] CPU: 0 PID: 1 Comm: swapper/0 Tainted: G        W       4.1.0-rc2-next-20150505-00344-g8577890defbf #79
	[    2.366682] Hardware name: NVIDIA Tegra SoC (Flattened Device Tree)
	[    2.372939] task: ee0d8b40 ti: ee0da000 task.ti: ee0da000
	[    2.378336] PC is at azx_bus_init+0x18/0xf4
	[    2.382516] LR is at hda_tegra_probe+0x6c/0x478
	[    2.387043] pc : [<c06156c4>]    lr : [<c061cf00>]    psr: 60000113
	[    2.387043] sp : ee0dbe38  ip : 00000000  fp : 00000000
	[    2.398501] r10: ed874c00  r9 : 000000fd  r8 : 00000000
	[    2.403717] r7 : ed874c10  r6 : 00000000  r5 : 00000000  r4 : ed016810
	[    2.410232] r3 : c08a2ad4  r2 : c08a1ea0  r1 : 00000000  r0 : ed016810
	[    2.416750] Flags: nZCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment kernel
	[    2.424046] Control: 10c5387d  Table: 8000406a  DAC: 00000015
	[    2.429783] Process swapper/0 (pid: 1, stack limit = 0xee0da210)
	[    2.435778] Stack: (0xee0dbe38 to 0xee0dc000)
	[    2.440129] be20:                                                       00000000 ed016810
	[    2.448297] be40: 00000000 c061cf00 00000000 ee0dbe5c ed8735d0 c0a7bc48 ed02fd50 ed016000
	[    2.456462] be60: c1250164 ed874c10 c0c66bf8 fffffdfb 00000000 000000fd c0b8dc98 c046664c
	[    2.464628] be80: c0466608 c1250164 ed874c10 00000000 c0c66bf8 c0464eb4 ed874c10 c0c66bf8
	[    2.472793] bea0: ed874c44 c0c43458 00000000 c04650d0 00000000 c0c66bf8 c046503c c04633b4
	[    2.480959] bec0: ee11bea4 ed85f390 c0c66bf8 ed017ac0 00000000 c0464634 c0ab2b7c c0c66bf8
	[    2.489125] bee0: c0bfde20 c0c66bf8 c0bfde20 ed01ce40 c0b7b414 c04656e8 c04665b0 c0bfde20
	[    2.497291] bf00: c0bfde20 c0009770 ee0d8b40 c0c02488 60000113 00000000 00000000 00000003
	[    2.505458] bf20: 00000000 c0c02488 60000113 00000000 c0b54598 c0b16a90 ef7fcc57 c0041228
	[    2.513624] bf40: c0a9150c ef7fcc5f 00000006 00000006 00000000 c0bf1fa8 c0bf2354 00000006
	[    2.521790] bf60: c0b8dc90 c0c7c000 c0c7c000 c0b8dc98 00000000 c0b54dd8 00000006 00000006
	[    2.529956] bf80: c0b54598 00000000 00000000 c07ff08c 00000000 00000000 00000000 00000000
	[    2.538122] bfa0: 00000000 c07ff094 00000000 c000f5a0 00000000 00000000 00000000 00000000
	[    2.546286] bfc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
	[    2.554451] bfe0: 00000000 00000000 00000000 00000000 00000013 00000000 fffff7ff c013f264
	[    2.562624] [<c06156c4>] (azx_bus_init) from [<c061cf00>] (hda_tegra_probe+0x6c/0x478)
	[    2.570535] [<c061cf00>] (hda_tegra_probe) from [<c046664c>] (platform_drv_probe+0x44/0xa4)
	[    2.578879] [<c046664c>] (platform_drv_probe) from [<c0464eb4>] (driver_probe_device+0x174/0x2b8)
	[    2.587739] [<c0464eb4>] (driver_probe_device) from [<c04650d0>] (__driver_attach+0x94/0x98)
	[    2.596172] [<c04650d0>] (__driver_attach) from [<c04633b4>] (bus_for_each_dev+0x6c/0xa0)
	[    2.604342] [<c04633b4>] (bus_for_each_dev) from [<c0464634>] (bus_add_driver+0x148/0x1f0)
	[    2.612597] [<c0464634>] (bus_add_driver) from [<c04656e8>] (driver_register+0x78/0xf8)
	[    2.620593] [<c04656e8>] (driver_register) from [<c0009770>] (do_one_initcall+0x8c/0x1d4)
	[    2.628765] [<c0009770>] (do_one_initcall) from [<c0b54dd8>] (kernel_init_freeable+0x144/0x1e4)
	[    2.637459] [<c0b54dd8>] (kernel_init_freeable) from [<c07ff094>] (kernel_init+0x8/0xe8)
	[    2.645543] [<c07ff094>] (kernel_init) from [<c000f5a0>] (ret_from_fork+0x14/0x34)

This is caused by azx_bus_init() trying to dereference chip->card, which
for the Tegra driver doesn't get initialized until sometime later during
the call to hda_tegra_create().

Fix this by mimicking the behaviour of the Intel driver and defer HDA
bus initialization until right before the call to snd_device_new().

Signed-off-by: Thierry Reding <treding at nvidia.com>
---
 sound/pci/hda/hda_tegra.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/sound/pci/hda/hda_tegra.c b/sound/pci/hda/hda_tegra.c
index 801e9fb4a467..db0bb50fa5b9 100644
--- a/sound/pci/hda/hda_tegra.c
+++ b/sound/pci/hda/hda_tegra.c
@@ -439,6 +439,10 @@ static int hda_tegra_create(struct snd_card *card,
 	chip->single_cmd = false;
 	chip->snoop = true;
 
+	err = azx_bus_init(chip, NULL, &hda_tegra_io_ops);
+	if (err < 0)
+		return err;
+
 	err = snd_device_new(card, SNDRV_DEV_LOWLEVEL, chip, &ops);
 	if (err < 0) {
 		dev_err(card->dev, "Error creating device\n");
@@ -475,10 +479,6 @@ static int hda_tegra_probe(struct platform_device *pdev)
 		return err;
 	}
 
-	err = azx_bus_init(chip, NULL, &hda_tegra_io_ops);
-	if (err < 0)
-		goto out_free;
-
 	err = hda_tegra_create(card, driver_flags, hda);
 	if (err < 0)
 		goto out_free;
-- 
2.3.5