[opensuse-kernel] USB devices inactive after kernel/stabel upgrade from 4.12.7-4 -> 4.12.8-2
after Leap 42.3 kernel/stable upgrade from 4.12.7-4 -> 4.12.8-2, what I'd initially thought was a problem with NVidia driver, appears to be a broader issue ... On 4 Leap machines, all linux/64, after upgrade, various USB devices are non-responsive once logged-in. On two machines WITH NVidia HW & driver, it's keyboard+mouse. The reason I initially thought it was X/Nvidia-related, is that the mouse/keyboard are live through grub menu, but were noticed dead at the SDDM login. On another, without any Nvidia HW or driver it's 2 wifi dongles. On the last, similarly without Nvidia, it's a USB-connected HD drive. In all 4 cases, switching back to 4.12.7-4 cleanly fixes the issues. Attempts to manually bring up the devices simply returns "device does not exist". I'm apparently not looking in the right place, cuz I see no logged *error* -- no OOPS, crashes, or even errors, yet. I can bisect if absolutely needed, but as these are OSB-pkg'd kernels, ideally, it's messy-ish. So, first question -- anyone else seeing similar problems? I can provide detail; Any specific pointers on what/how to debug/trace the problem? -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Am Freitag, den 18.08.2017, 10:00 -0700 schrieb PGNet Dev:
after Leap 42.3 kernel/stable upgrade from 4.12.7-4 -> 4.12.8-2, what I'd initially thought was a problem with NVidia driver, appears to be a broader issue ...
On 4 Leap machines, all linux/64, after upgrade, various USB devices are non-responsive once logged-in.
On two machines WITH NVidia HW & driver, it's keyboard+mouse. The reason I initially thought it was X/Nvidia-related, is that the mouse/keyboard are live through grub menu, but were noticed dead at the SDDM login.
On another, without any Nvidia HW or driver it's 2 wifi dongles.
On the last, similarly without Nvidia, it's a USB-connected HD drive.
In all 4 cases, switching back to 4.12.7-4 cleanly fixes the issues.
Hi, could you provide dmesg, ideally of both versions? Regards Oliver -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Hi,
could you provide dmesg, ideally of both versions?
Regards Oliver
sure. for a leap 42.3 desktop, for which the fail == no mouse/keyboard response boot 4.12.7, https://pastebin.com/xJg5H3Bs , keyboard & mouse functional boot 4.12.8, https://pastebin.com/VGm3ua1F , keyboard & mouse non-responsive -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Am Sonntag, den 20.08.2017, 10:36 -0700 schrieb PGNet Dev:
Hi,
could you provide dmesg, ideally of both versions?
Regards Oliver
sure.
for a leap 42.3 desktop, for which the fail == no mouse/keyboard response
boot 4.12.7, https://pastebin.com/xJg5H3Bs , keyboard & mouse functional boot 4.12.8, https://pastebin.com/VGm3ua1F , keyboard & mouse non-responsive
Hi, the obvious change in the logs is that the failing version does not log an application of the AMD freeze work around from here 0a99e8ac430a2 (Shane Huang 2008-11-25 15:12:33 +0800 216) /* SB600 and old version of SB700 have a bug in EHCI controller, b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 217) * which causes usb devices lose response in some cases. b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 218) */ 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 219) if ((pdev->device == 0x4386 || pdev->device == 0x4396) && 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 220) usb_amd_hang_symptom_quirk()) { 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 221) u8 tmp; 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 222) ehci_info(ehci, "applying AMD SB600/SB700 USB freeze workaround\n"); 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 223) pci_read_config_byte(pdev, 0x53, &tmp); 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 224) pci_write_config_byte(pdev, 0x53, tmp | (1<<3)); b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 225) } b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 226) break; This code however has not changed in four years, so I do not understand what is happening. Thus I would prefer you to validate the theory. Can you make a test kernel which applies the work around unconditionally or do you want me to have the build service make a test kernel? Regards Oliver -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 8/21/17 3:06 AM, Oliver Neukum wrote:
the obvious change in the logs is that the failing version does not log an application of the AMD freeze work around from here
0a99e8ac430a2 (Shane Huang 2008-11-25 15:12:33 +0800 216) /* SB600 and old version of SB700 have a bug in EHCI controller, b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 217) * which causes usb devices lose response in some cases. b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 218) */ 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 219) if ((pdev->device == 0x4386 || pdev->device == 0x4396) && 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 220) usb_amd_hang_symptom_quirk()) { 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 221) u8 tmp; 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 222) ehci_info(ehci, "applying AMD SB600/SB700 USB freeze workaround\n"); 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 223) pci_read_config_byte(pdev, 0x53, &tmp); 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 224) pci_write_config_byte(pdev, 0x53, tmp | (1<<3)); b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 225) } b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 226) break;
This code however has not changed in four years, so I do not understand what is happening. Thus I would prefer you to validate the theory. Can you make a test kernel which applies the work around unconditionally or do you want me to have the build service make a test kernel?
Happy to test. If you can whip up a test kernel pkg without too much trouble, that'll make it simpler on my end, and I can test immediately. -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Am Montag, den 21.08.2017, 06:07 -0700 schrieb PGNet Dev:
On 8/21/17 3:06 AM, Oliver Neukum wrote:
the obvious change in the logs is that the failing version does not log an application of the AMD freeze work around from here
0a99e8ac430a2 (Shane Huang 2008-11-25 15:12:33 +0800 216) /* SB600 and old version of SB700 have a bug in EHCI controller, b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 217) * which causes usb devices lose response in some cases. b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 218) */ 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 219) if ((pdev->device == 0x4386 || pdev->device == 0x4396) && 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 220) usb_amd_hang_symptom_quirk()) { 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 221) u8 tmp; 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 222) ehci_info(ehci, "applying AMD SB600/SB700 USB freeze workaround\n"); 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 223) pci_read_config_byte(pdev, 0x53, &tmp); 3ad145b62a15c (Huang Rui 2013-10-03 23:37:12 +0800 224) pci_write_config_byte(pdev, 0x53, tmp | (1<<3)); b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 225) } b09bc6cbae4dd (Andiry Xu 2008-11-14 11:42:29 +0800 226) break;
This code however has not changed in four years, so I do not understand what is happening. Thus I would prefer you to validate the theory. Can you make a test kernel which applies the work around unconditionally or do you want me to have the build service make a test kernel?
Happy to test.
If you can whip up a test kernel pkg without too much trouble, that'll make it simpler on my end, and I can test immediately.
The test kernel is being built https://build.opensuse.org/project/monitor?project=home%3Aoneukum%3Aamd_free... HTH Oliver -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
The test kernel is being built https://build.opensuse.org/project/monitor?project=home%3Aoneukum%3Aamd_free...
HTH
Thx, it does. FYI, the build's almost there ... all pkgs for x86_64 built, except https://build.opensuse.org/package/live_build_log/home:oneukum:amd_freeze_te... which, at least from my view, has been stuck @ [ 4927s] CC [M] drivers/staging/speakup/keyhelp.o for an hour+ ... -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 8/21/17 10:49 AM, PGNet Dev wrote:
The test kernel is being built https://build.opensuse.org/project/monitor?project=home%3Aoneukum%3Aamd_free...
Your test kernel FIXES the problem ... With kernel from your test-build zypper se -s | grep -i kernel-default | grep ^i | grep 4.12.8 i+ | kernel-default | package | 4.12.8-2.1.g4d7933a | x86_64 | TEST-KernelStable i+ | kernel-default-devel | package | 4.12.8-2.1.g4d7933a | x86_64 | TEST-KernelStable uname -r 4.12.8-2.g4d7933a-default keyboard and mouse *ARE*, now, fully functional.
From before
boot 4.12.7, https://pastebin.com/xJg5H3Bs , keyboard & mouse functional boot 4.12.8, https://pastebin.com/VGm3ua1F , keyboard & mouse non-responsive With your fix boot 4.12.8 'amd freeze', https://pastebin.com/cns5uXkJ , keyboard & mouse functional -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Am Montag, den 21.08.2017, 15:34 -0700 schrieb PGNet Dev:
From before
boot 4.12.7, https://pastebin.com/xJg5H3Bs , keyboard & mouse functional boot 4.12.8, https://pastebin.com/VGm3ua1F , keyboard & mouse non-responsive
With your fix
boot 4.12.8 'amd freeze', https://pastebin.com/cns5uXkJ , keyboard & mouse functional
Hi, very well, we now know what is going wrong, but not why. While I am looking at this, could you please make a bugzilla report against Tumbleweed and assign it to me? Regards Oliver -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 8/22/17 7:35 AM, Oliver Neukum wrote:
very well, we now know what is going wrong, but not why. While I am looking at this, could you please make a bugzilla report against Tumbleweed and assign it to me?
sure. -> https://bugzilla.opensuse.org/show_bug.cgi?id=1055044 -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
participants (2)
-
Oliver Neukum -
PGNet Dev