gspca_ov534: kernel oops when connecting Hercules Blog Webcam

Hi!

After update from 3.5 kernel to newer version I got kernel oops when I connect my Hercules Blog Webcam. The full error stacktrace is at the end of this e-mail.

Commit bisect revealed the regression at:
-------------------------
1bd7d6adc691993206cf7dd69f1aaf8dccb06677 is the first bad commit commit 1bd7d6adc691993206cf7dd69f1aaf8dccb06677
Author: Antonio Ospite <ospite@xxxxxxxxxxxx>
Date: Wed May 16 18:42:46 2012 -0300

[media] gspca_ov534: Convert to the control framework

:040000 040000 81bb6d86a59d2fca15fea4d43a8abe34354cf69e 6b7c2077ae5d8bdea32864841b5cd14149c6a44a M drivers
-------------------------

I also try to change the gspca_ov534 module source (ov534.c) from the 3.8 kernel and added some additional null checks in sd_start function and add the extensive logging to it. With those changes my webcam is working OK. As far as I can see, sd_start function is called couple of times, but when it is called the first time (when a usb cable was inserted), sd struct is almost empty and without null check this leads to kernel oops. Here is the part of test version log when sd_start was called first time:

sd_start: NO sd->hue!
sd_start: NO sd->saturation!
sd_start: NO sd->autogain!
sd_start: NO sd->autowhitebalance!
sd_start: NO sd->autoexposure!
sd_start: NO sd->gain!
sd_start: NO sd->exposure!
sd_start: NO sd->brightness!
sd_start: NO sd->contrast!
sd_start: NO sd->sharpness!
sd_start: NO sd->hflip and sd->vflip!
sd_start: NO sd->plfreq!

Additional info can be found here: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1173723/

Here is the error stacktrace:
kernel: [ 52.679705] usb 3-2: new high-speed USB device number 2 using xhci_hcd
kernel: [ 52.697906] usb 3-2: New USB device found, idVendor=06f8, idProduct=3002
kernel: [ 52.697910] usb 3-2: New USB device strings: Mfr=1, Product=2, SerialNumber=0
kernel: [ 52.697912] usb 3-2: Product: Hercules Blog Microphone
kernel: [ 52.697914] usb 3-2: Manufacturer: Hercules Blog Webcam
kernel: [ 52.708983] Linux video capture interface: v2.00
kernel: [ 52.710778] gspca_main: v2.14.0 registered
kernel: [ 52.712210] gspca_main: ov534-2.14.0 probing 06f8:3002
kernel: [ 55.506311] BUG: unable to handle kernel NULL pointer dereference at 0000000000000050
kernel: [ 55.506367] IP: [<ffffffffa03c1b01>] v4l2_ctrl_g_ctrl+0x11/0x60 [videodev]
kernel: [ 55.506414] PGD 0
kernel: [ 55.506429] Oops: 0000 [#1] SMP
kernel: [ 55.506453] Modules linked in: gspca_ov534(+) gspca_main videodev rfcomm bnep ppdev bluetooth binfmt_misc snd_hda_codec_hdmi snd_hda_codec_realtek stir4200 irda crc_ccitt usblp snd_hda_intel snd_hda_codec snd_hwdep snd_pcm hid_generic snd_page_alloc snd_seq_midi snd_seq_midi_event usbhid snd_rawmidi snd_seq snd_seq_device snd_timer hid i915 snd psmouse drm_kms_helper serio_raw mei_me drm mei soundcore video i2c_algo_bit lpc_ich mac_hid coretemp lp parport firewire_ohci firewire_core crc_itu_t ahci libahci alx mdio r8169 mii [last unloaded: parport_pc]
kernel: [ 55.506819] CPU: 3 PID: 4352 Comm: modprobe Not tainted 3.11.0-031100rc2-generic #201307211535
kernel: [ 55.506864] Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./Z77-DS3H, BIOS F9 09/19/2012
kernel: [ 55.506913] task: ffff8801c20f9770 ti: ffff8801ceaa0000 task.ti: ffff8801ceaa0000
kernel: [ 55.506952] RIP: 0010:[<ffffffffa03c1b01>] [<ffffffffa03c1b01>] v4l2_ctrl_g_ctrl+0x11/0x60 [videodev]
kernel: [ 55.507005] RSP: 0018:ffff8801ceaa1af8 EFLAGS: 00010292
kernel: [ 55.507033] RAX: 0000000000000001 RBX: 0000000000000000 RCX: 000000000001988b
kernel: [ 55.507069] RDX: 000000000001988a RSI: ffffffffa032745a RDI: 0000000000000000
kernel: [ 55.507106] RBP: ffff8801ceaa1b28 R08: 0000000000017380 R09: ffffea0008419d80
kernel: [ 55.507142] R10: ffffffff81538f5a R11: 0000000000000002 R12: ffffffffa03273dc
kernel: [ 55.507178] R13: ffffffffa03273dc R14: 0000000000000000 R15: ffffffffa03270a0
kernel: [ 55.507215] FS: 00007f72d564a740(0000) GS:ffff88021f380000(0000) knlGS:0000000000000000
kernel: [ 55.507256] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
kernel: [ 55.507286] CR2: 0000000000000050 CR3: 00000001bd1f0000 CR4: 00000000001407e0
kernel: [ 55.507323] Stack:
kernel: [ 55.507335] ffff8801ceaa1b28 ffffffffa0325cff ffff8801000001f4 ffff8801ceb44000
kernel: [ 55.507380] ffffffffa03273dc ffff8801ceb44000 ffff8801ceaa1b58 ffffffffa032688e
kernel: [ 55.507426] ffff8801ceb44000 ffffffffa03274f0 ffffffffa03274f0 ffff8801ceb44380
kernel: [ 55.507471] Call Trace:
kernel: [ 55.507490] [<ffffffffa0325cff>] ? sccb_w_array+0x3f/0x80 [gspca_ov534]
kernel: [ 55.507527] [<ffffffffa032688e>] sd_start+0xce/0x2b0 [gspca_ov534]
kernel: [ 55.507561] [<ffffffffa0326bf9>] sd_init+0x189/0x1e8 [gspca_ov534]
kernel: [ 55.507596] [<ffffffffa02a0c95>] gspca_dev_probe2+0x285/0x410 [gspca_main]
kernel: [ 55.507634] [<ffffffffa02a0e58>] gspca_dev_probe+0x38/0x60 [gspca_main]
kernel: [ 55.507670] [<ffffffffa0325081>] sd_probe+0x21/0x30 [gspca_ov534]
kernel: [ 55.507706] [<ffffffff8153c960>] usb_probe_interface+0x1c0/0x2f0
kernel: [ 55.507740] [<ffffffff8148758c>] really_probe+0x6c/0x330
kernel: [ 55.507771] [<ffffffff814879d7>] driver_probe_device+0x47/0xa0
kernel: [ 55.507803] [<ffffffff81487adb>] __driver_attach+0xab/0xb0
kernel: [ 55.507834] [<ffffffff81487a30>] ? driver_probe_device+0xa0/0xa0
kernel: [ 55.507867] [<ffffffff814857be>] bus_for_each_dev+0x5e/0x90
kernel: [ 55.507899] [<ffffffff8148714e>] driver_attach+0x1e/0x20
kernel: [ 55.507929] [<ffffffff81486bdc>] bus_add_driver+0x10c/0x290
kernel: [ 55.507961] [<ffffffff8148805d>] driver_register+0x7d/0x160
kernel: [ 55.507993] [<ffffffff8153b590>] usb_register_driver+0xa0/0x160
kernel: [ 55.508027] [<ffffffffa0067000>] ? 0xffffffffa0066fff
kernel: [ 55.508056] [<ffffffffa006701e>] sd_driver_init+0x1e/0x1000 [gspca_ov534]
kernel: [ 55.508094] [<ffffffff8100212a>] do_one_initcall+0xfa/0x1b0
kernel: [ 55.508126] [<ffffffff810578c3>] ? set_memory_nx+0x43/0x50
kernel: [ 55.508160] [<ffffffff81712e8d>] do_init_module+0x80/0x1d1
kernel: [ 55.508193] [<ffffffff810d2079>] load_module+0x4c9/0x5f0
kernel: [ 55.508223] [<ffffffff810cf7b0>] ? add_kallsyms+0x210/0x210
kernel: [ 55.508254] [<ffffffff810d2254>] SyS_init_module+0xb4/0x100
kernel: [ 55.508286] [<ffffffff817333ef>] tracesys+0xe1/0xe6
kernel: [ 55.508312] Code: a0 09 00 00 48 c7 c7 30 c3 3c a0 e8 7a 38 ca e0 eb cf 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 89 e5 53 48 89 fb 48 83 ec 28 <8b> 47 50 83 e8 05 83 f8 02 77 09 80 b8 20 8c 3c a0 00 74 1d 48
kernel: [ 55.508559] RIP [<ffffffffa03c1b01>] v4l2_ctrl_g_ctrl+0x11/0x60 [videodev]
kernel: [ 55.510605] RSP <ffff8801ceaa1af8>
kernel: [ 55.512641] CR2: 0000000000000050
kernel: [ 55.525424] ---[ end trace 6786f15abfd2ac90 ]---

Regards,
 Yaroslav

Hi Yaroslav, I'll try to take a look this week-end.

[...]

I saw you also tested with 3.11-rc2 kernel, and the issue is still
there: https://launchpadlibrarian.net/145608306/kern.log

Thanks,
Antonio

Antonio Ospite
http://ao2.it

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?

---

Hi Yaroslav,

the patch below should fix the Oops caused by sd_start() called too early, but I am not sure about why sd_start() was called from sd_init() for Hercules webcams in the first place, maybe the snippet marked with:

/* (from ms-win trace) */

in sd_start() must be moved to sd_init() too.

Let me know if the change below alone is enough and the webcam keeps working, a test with suspend and resume would good to have too.

Thanks,
Antonio

 drivers/media/usb/gspca/ov534.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

index 2e28c81..03a33c4 100644
@@ -1305,8 +1305,7 @@ static int sd_init(struct gspca_dev *gspca_dev)
 	ov534_set_led(gspca_dev, 1);
 	sccb_w_array(gspca_dev, sensor_init[sd->sensor].val,
 			sensor_init[sd->sensor].len);
-	if (sd->sensor == SENSOR_OV767x)
-		sd_start(gspca_dev);
+
 	sd_stopN(gspca_dev);
 /*	set_frame_rate(gspca_dev);	*/
 

1.8.4.rc1

Hi Antonio,

I've tested your patch with the latest kernel (3.11.0-rc4) - the webcam works OK. After suspend and resume, the webcam works OK too.

Bye, Yaroslav

sd_start() operates on device controls but after the conversion to the v4l2 control framework in commits 62bba5d and 1bd7d6a controls are initialized in sd_init_controls() which is called _after_ sd_init():

The change fixes a NULL pointer dereference for Hercules Blog Webcam; the problem is observable since 3.6:

gspca_main: v2.14.0 registered
gspca_main: ov534-2.14.0 probing 06f8:3002

  BUG: unable to handle kernel NULL pointer dereference at 0000000000000050
  IP: [<ffffffffa03c1b01>] v4l2_ctrl_g_ctrl+0x11/0x60 [videodev]
  PGD 0
  Oops: 0000 [#1] SMP
  Modules linked in: gspca_ov534(+) gspca_main videodev rfcomm bnep ppdev bluetooth binfmt_misc snd_hda_codec_hdmi snd_hda_codec_realtek stir4200 irda crc_ccitt usblp snd_hda_intel snd_hda_codec snd_hwdep snd_pcm hid_generic snd_page_alloc snd_seq_midi snd_seq_midi_event usbhid snd_rawmidi snd_seq snd_seq_device snd_timer hid i915 snd psmouse drm_kms_helper serio_raw mei_me drm mei soundcore video i2c_algo_bit lpc_ich mac_hid coretemp lp parport firewire_ohci firewire_core crc_itu_t ahci libahci alx mdio r8169 mii [last unloaded: parport_pc]
  CPU: 3 PID: 4352 Comm: modprobe Not tainted 3.11.0-031100rc2-generic #201307211535
  Hardware name: Gigabyte Technology Co., Ltd. To be filled by O.E.M./Z77-DS3H, BIOS F9 09/19/2012
  task: ffff8801c20f9770 ti: ffff8801ceaa0000 task.ti: ffff8801ceaa0000
  RIP: 0010:[<ffffffffa03c1b01>]  [<ffffffffa03c1b01>] v4l2_ctrl_g_ctrl+0x11/0x60 [videodev]
  RSP: 0018:ffff8801ceaa1af8  EFLAGS: 00010292
  RAX: 0000000000000001 RBX: 0000000000000000 RCX: 000000000001988b
  RDX: 000000000001988a RSI: ffffffffa032745a RDI: 0000000000000000
  RBP: ffff8801ceaa1b28 R08: 0000000000017380 R09: ffffea0008419d80
  R10: ffffffff81538f5a R11: 0000000000000002 R12: ffffffffa03273dc
  R13: ffffffffa03273dc R14: 0000000000000000 R15: ffffffffa03270a0
  FS:  00007f72d564a740(0000) GS:ffff88021f380000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000000000050 CR3: 00000001bd1f0000 CR4: 00000000001407e0
  Stack:
   ffff8801ceaa1b28 ffffffffa0325cff ffff8801000001f4 ffff8801ceb44000
   ffffffffa03273dc ffff8801ceb44000 ffff8801ceaa1b58 ffffffffa032688e
   ffff8801ceb44000 ffffffffa03274f0 ffffffffa03274f0 ffff8801ceb44380
  Call Trace:
   [<ffffffffa0325cff>] ? sccb_w_array+0x3f/0x80 [gspca_ov534]
   [<ffffffffa032688e>] sd_start+0xce/0x2b0 [gspca_ov534]
   [<ffffffffa0326bf9>] sd_init+0x189/0x1e8 [gspca_ov534]
   [<ffffffffa02a0c95>] gspca_dev_probe2+0x285/0x410 [gspca_main]
   [<ffffffffa02a0e58>] gspca_dev_probe+0x38/0x60 [gspca_main]
   [<ffffffffa0325081>] sd_probe+0x21/0x30 [gspca_ov534]
   [<ffffffff8153c960>] usb_probe_interface+0x1c0/0x2f0
   [<ffffffff8148758c>] really_probe+0x6c/0x330
   [<ffffffff814879d7>] driver_probe_device+0x47/0xa0
   [<ffffffff81487adb>] __driver_attach+0xab/0xb0
   [<ffffffff81487a30>] ? driver_probe_device+0xa0/0xa0
   [<ffffffff814857be>] bus_for_each_dev+0x5e/0x90
   [<ffffffff8148714e>] driver_attach+0x1e/0x20
   [<ffffffff81486bdc>] bus_add_driver+0x10c/0x290
   [<ffffffff8148805d>] driver_register+0x7d/0x160
   [<ffffffff8153b590>] usb_register_driver+0xa0/0x160
   [<ffffffffa0067000>] ? 0xffffffffa0066fff
   [<ffffffffa006701e>] sd_driver_init+0x1e/0x1000 [gspca_ov534]
   [<ffffffff8100212a>] do_one_initcall+0xfa/0x1b0
   [<ffffffff810578c3>] ? set_memory_nx+0x43/0x50
   [<ffffffff81712e8d>] do_init_module+0x80/0x1d1
   [<ffffffff810d2079>] load_module+0x4c9/0x5f0
   [<ffffffff810cf7b0>] ? add_kallsyms+0x210/0x210
   [<ffffffff810d2254>] SyS_init_module+0xb4/0x100
   [<ffffffff817333ef>] tracesys+0xe1/0xe6
  Code: a0 09 00 00 48 c7 c7 30 c3 3c a0 e8 7a 38 ca e0 eb cf 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 89 e5 53 48 89 fb 48 83 ec 28 <8b> 47 50 83 e8 05 83 f8 02 77 09 80 b8 20 8c 3c a0 00 74 1d 48
  RIP  [<ffffffffa03c1b01>] v4l2_ctrl_g_ctrl+0x11/0x60 [videodev]
   RSP <ffff8801ceaa1af8>
  CR2: 0000000000000050
  ---[ end trace 6786f15abfd2ac90 ]---

Original bug report from:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1173723/

---
 drivers/media/usb/gspca/ov534.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

index 2e28c81..03a33c4 100644
@@ -1305,8 +1305,7 @@ static int sd_init(struct gspca_dev *gspca_dev)
 	ov534_set_led(gspca_dev, 1);
 	sccb_w_array(gspca_dev, sensor_init[sd->sensor].val,
 			sensor_init[sd->sensor].len);
-	if (sd->sensor == SENSOR_OV767x)
-		sd_start(gspca_dev);
+
 	sd_stopN(gspca_dev);
 /*	set_frame_rate(gspca_dev);	*/
 

1.8.4.rc2

Hi,

Thanks for the patch I've added this to my "gspca" tree, and this will be included in my next pull-request to Mauro for 3.12

Regards,

Hans

Thanks HdG.

It's fine with me to have the patch in 3.12 and then have it picked up for inclusion in stable releases, I was just wondering why you didn't consider it as a fix for 3.11, the patch fixes an actual crash experienced by a user.

Regards,
Antonio

Antonio Ospite
http://ao2.it

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?

Hi,

I did not have time to do v4l work before now, and atm it is simply too late for 3.11

Regards,

Hans