Bug 60131
Summary: | [ivb dp/hdmi] Problem with detecting HDMI plugging/unplugging (Samsung TV, Intel card) | ||
---|---|---|---|
Product: | Drivers | Reporter: | Robert Gonciarz (gonciarz) |
Component: | Video(DRI - Intel) | Assignee: | Antti Koskipaa (antti.koskipaa) |
Status: | RESOLVED CODE_FIX | ||
Severity: | normal | CC: | daniel, intel-gfx-bugs |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | 3.9.7 | Subsystem: | |
Regression: | No | Bisected commit-id: | |
Attachments: |
dmesg_hdmi_off
dmesg_hdmi_on dmesg_dp_off dmesg_dp_on Kernel's config dmesg_3.1.10_0_hdmioff_dp0off dmesg_3.1.10_1_hdmion_dp0off dmesg_3.1.10_2_hdmioff_dp0off dmesg_3.1.10_3_hdmioff_dp0on dmesg_3.2.0-3-amd64_0_hdmioff_dp0off dmesg_3.2.0-3-amd64_1_hdmion_dp0off dmesg_3.2.0-3-amd64_2_hdmioff_dp0off dmesg_3.2.0-3-amd64_3_hdmioff_dp0on 3.12.0-rc4_0_hdmioff_dp0off 3.12.0-rc4_1_hdmion_dp0off 3.12.0-rc4_2_hdmioff_dp0off 3.12.0-rc4_3_hdmioff_dp0on |
Description
Robert Gonciarz
2013-06-25 21:21:51 UTC
A few questions: - With some quick googling it sounds like your board has both a HDMI and a DP port. Do both ports not work? - Please boot with drm.debug=0xe added to your kernel cmdline, attach the screens once the system has booted and then grab the complete dmesg and attach it to this bug. - Can you please test the latest drm-intel-nightly branch from http://cgit.freedesktop.org/~danvet/drm-intel/ to check that we haven't fixed this bug already? One more: You're saying hotplug detection worked before, so this is a regression, right? Hi Daniel, I'm sure HDMI was working correctly, but I don't remember if I have ever checked hotplugging. So I'm not sure if it's better to rename the subject because even if I boot with HDMI already plugged, xserver shows it can't detect any screen. On Wed, 26 Jun 2013 09:41:57 +0000 (UTC) bugzilla-daemon@bugzilla.kernel.org wrote: > https://bugzilla.kernel.org/show_bug.cgi?id=60131 > > > Daniel Vetter <daniel@ffwll.ch> changed: > > What |Removed |Added > ---------------------------------------------------------------------------- > Regression|No |Yes > > > > > --- Comment #2 from Daniel Vetter <daniel@ffwll.ch> 2013-06-26 09:41:57 --- > One more: You're saying hotplug detection worked before, so this is a > regression, right? > Created attachment 106171 [details]
dmesg_hdmi_off
Created attachment 106181 [details]
dmesg_hdmi_on
Created attachment 106191 [details]
dmesg_dp_off
Created attachment 106201 [details]
dmesg_dp_on
Created attachment 106211 [details]
Kernel's config
Hi, I've checked out the sources from the repo git clone git://people.freedesktop.org/~danvet/drm-intel -b drm-intel-nightly that's the revision: f3478bb (Date: Thu Jun 27 13:30:41 2013 +0200) Then I copied my .config file into kernel source make menuconfig fakeroot make-kpkg -j4 --initrd kernel_image kernel_headers dpkg -i ../*deb Set cmdline in /etc/default/grub GRUB_CMDLINE_LINUX_DEFAULT="quiet drm.debug=0xe" update-grub reboot after boot: uname -a Linux smithy 3.10.0-rc7+ #1 SMP Thu Jun 27 20:07:42 CEST 2013 x86_64 GNU/Linux cat /proc/cmdline BOOT_IMAGE=/vmlinuz-3.10.0-rc7+ root=UUID=1acb149f-252a-4c38-a0f5-81fc4d3d2b66 ro quiet drm.debug=0xe Then I did the following sequence: 0. Turned on Monitor and set the mode for HDMI1 + connected HDMI cable 1. plugged HDMI cable 2. dmesg -l debug > dmesg_hdmi_on 3. unplagged HDMI cable 4. dmesg -l debug > dmesg_hdmi_off 5. plugged DP adapter/cable 6. dmesg -l debug > dmesg_dp_on 7. unplagged DP adapter/cable 8. dmesg -l debug > dmesg_dp_off The files should be in the attachments. If you still need more info, please let me know. Best regards Robert I forgot to add. Yes this mainboard has 3 graphic ports: VGA/D-SUB, HDMI and DP. I've checked udevadm monitor before and after plugging/unplugging HDMI/DP cable. And I haven't seen any new info. It looks like dmesg registered something when I shuffled with DP. R. Hi Daniel is there any other info that I may provide for you? Best regards R. Can you please digg out a dmesg with drm.debug=0xe set from an old working kernel? Afaics for the first port we simply don't get any hotplug interrupts and for the 2nd one the device doesn't really seem to exists ... Hi, 1. The problem is I don't remember exactly which kernel was used when monitor was properly recognized by HDMI/DP. I've checked several kernels (from official kernel.org and using debian's one.) and none of them was working: - vmlinuz-2.6.39.4 - vmlinuz-3.1.10 - vmlinuz-3.2.0-3-amd64 - vmlinuz-3.2.48 (?) 2. But I've discovered a difference in dmesg output when plugging/unplugging hdmi cable if we compare 3.1.10 and 3.2.0-3-amd64 (debian's) kernels. Scenario is the same as previously: 0. Setup TV (hdmi) and boot the system + dmesg 1. plug hdmi + dmesg 2. unplug hdmi + dmesg 3. plug dp + dmesg 4. unplug dp + dmesg #Size output 72089 dmesg_3.1.10_0_hdmioff_dp0off 85601 dmesg_3.1.10_1_hdmion_dp0off 85601 dmesg_3.1.10_2_hdmioff_dp0off 86926 dmesg_3.1.10_3_hdmioff_dp0on 87276 dmesg_3.2.0-3-amd64_0_hdmioff_dp0off 87276 dmesg_3.2.0-3-amd64_1_hdmion_dp0off 87276 dmesg_3.2.0-3-amd64_2_hdmioff_dp0off 88601 dmesg_3.2.0-3-amd64_3_hdmioff_dp0on Created attachment 106966 [details]
dmesg_3.1.10_0_hdmioff_dp0off
Created attachment 106967 [details]
dmesg_3.1.10_1_hdmion_dp0off
Created attachment 106968 [details]
dmesg_3.1.10_2_hdmioff_dp0off
Created attachment 106969 [details]
dmesg_3.1.10_3_hdmioff_dp0on
Created attachment 106970 [details]
dmesg_3.2.0-3-amd64_0_hdmioff_dp0off
Created attachment 106971 [details]
dmesg_3.2.0-3-amd64_1_hdmion_dp0off
Created attachment 106972 [details]
dmesg_3.2.0-3-amd64_2_hdmioff_dp0off
Created attachment 106973 [details]
dmesg_3.2.0-3-amd64_3_hdmioff_dp0on
Is there anything else I may do to speed up the fixing process? Best regards R. (In reply to Robert Gonciarz from comment #22) > Is there anything else I may do to speed up the fixing process? Sorry, this bug seems to have been neglected a bit. There have been some hotplug related fixes lately though, so could you give the latest drm-intel-nightly a try, please? Hi, sure, no problem. Thanks for the reply. I may check the nightly changes, but I'll do that rather during the weekend. I'll also try to isolate which kernel version contains the change that has broken this bug. My server has Core i7 processor so compiling new 'eggs' shouldn't be a problem. Hopefully I may be able to connect the monitor to my server and I also hope you may guide me how to do that properly. I believe that fixing drm is only the first step. Honestly I was even thinking of buying Raspberry Pi to work around this issue. Best regards Robert Sorry for delay. I've check-out the HEAD yesterday (commit b7214fceac0659d440adfa5f1acec238fb7217fc, Sun Oct 13 21:27:35 2013 +0200), built the kernel, deployed it and booted. uname -r: 3.12.0-rc4+ but it seems that right now neither DP nor HDMI doesn't detect TV/Monitor at all. I'll try to compare those kernel versions as well: linux-3.2 linux-3.2.1 linux-3.2.10 linux-3.2.20 linux-3.2.30 linux-3.2.40 maybe it will be easier to limit the scope of the searching. Created attachment 110971 [details]
3.12.0-rc4_0_hdmioff_dp0off
Created attachment 110981 [details]
3.12.0-rc4_1_hdmion_dp0off
Created attachment 110991 [details]
3.12.0-rc4_2_hdmioff_dp0off
Created attachment 111001 [details]
3.12.0-rc4_3_hdmioff_dp0on
(In reply to Robert Gonciarz from comment #25) > uname -r: > 3.12.0-rc4+ > > but it seems that right now neither DP nor HDMI doesn't detect TV/Monitor at > all. That's correct. I presume you've tried unplugging VGA, and plugging/unplugging DP and HDMI, at the motherboard end? With TV on. If you boot with *only* DP or HDMI connected (unplug VGA, try both, one at a time) does the BIOS screen show up? > I'll try to compare those kernel versions as well: > linux-3.2 > linux-3.2.1 > linux-3.2.10 > linux-3.2.20 > linux-3.2.30 > linux-3.2.40 > maybe it will be easier to limit the scope of the searching. In this case it might indeed be useful to find a working kernel, if you say there was one. Is this still happening/relevant? Marking as a non-regression because an older kernel could not be found that doesn't have this issue. Presuming fixed with latest upstream, please reopen if that's not the case and supply the requested information. |