Summary: | Multiple cards with modesetting results in Segmentation Fault | ||
---|---|---|---|
Product: | xorg | Reporter: | ToBeReplaced |
Component: | Driver/AMDgpu | Assignee: | xf86-video-ati maintainers <xorg-driver-ati> |
Status: | RESOLVED NOTOURBUG | QA Contact: | Xorg Project Team <xorg-team> |
Severity: | normal | ||
Priority: | medium | ||
Version: | unspecified | ||
Hardware: | x86-64 (AMD64) | ||
OS: | Linux (All) | ||
Whiteboard: | |||
i915 platform: | i915 features: |
Description
ToBeReplaced
2019-02-19 22:05:15 UTC
Whoops, forgot to add that I attempted to receive support through #xorg on freenode (no responses), then later at https://superuser.com/questions/1406211/xorg-multi-card-modesetting-configuration-segmentation-fault (no answers). If this is not the appropriate channel for escalation please let me know where would be more appropriate. modesetting driver issues need to be reported at https://gitlab.freedesktop.org/xorg/xserver/issues/new . That said, does removing the Screen stanzas from Section "Device" help? That is for a so-called "Zaphod" configuration with multiple "Device" and "Screen" sections for the same GPU. Thank you for the quick reply. I'll post an issue there, then close this one. Removing the Screen sections does not help, unfortunately. I'll post another log in a bit. The end goal is to use it with ZaphodHeads actually, so when moving the issue I'll include the ZaphodHeads lines and a log with those included. (In reply to ToBeReplaced from comment #3) > Thank you for the quick reply. I'll post an issue there, then close this one. This one's already resolved, no need to close it. > Removing the Screen sections does not help, unfortunately. That's not what I mean. I mean the Screen 0/1 stanzas in Section "Device". > The end goal is to use it with ZaphodHeads actually, so when moving the issue > I'll include the ZaphodHeads lines and a log with those included. Like the Screen stanza, Option "ZaphodHeads" is (as the name suggests) only relevant with a Zaphod configuration, not with a single screen per GPU as in your case. Thank you for all of the help. I have come to a resolution based on your hunch about removing Screen lines in the device sections. The segmentation fault appears to occur if there is no "Screen 0" for a given BusID. The core misunderstanding was that the screen number in the ServerLayout section is *not* related to the screen number in the Device section. My comment about ZaphodHeads was because I originally had a third screen based off of the second bus id, but I removed it from the configuration since it was not required to cause a Segmentation Fault. My end goal was display :0.0 on one card, :0.1 on another card, and :0.2 on that second card. A working configuration, with comments for the "surprises" I had is below, in case it is useful for anyone in the future or for improving existing documentaiton. Section "ServerLayout" Identifier "Example" Screen 0 "A" Screen 1 "B" Screen 2 "C" EndSection Section "Device" Identifier "A Card" Driver "modesetting" Screen 0 BusID "PCI:16@0:0:0" Option "kmsdev" "/dev/dri/card1" Option "Monitor-DVI-D-1" EndSection Section "Monitor" Identifier "A Monitor" EndSection Section "Screen" Identifier "A" Device "A Card" Monitor "A Monitor" EndSection Section "Device" Identifier "B Card" Driver "modesetting" Screen 0 # 1st screen for bus, 2nd screen in layout BusID "PCI:56@0:0:0" Option "kmsdev" "/dev/dri/card0" Option "Monitor-HDMI-1" EndSection Section "Monitor" Identifier "B Monitor" EndSection Section "Screen" Identifier "B" Device "B Card" Monitor "B Monitor" EndSection Section "Device" Identifier "C Card" Driver "modesetting" Screen 1 # 2nd screen for bus, 3rd screen in layout BusID "PCI:56@0:0:0" Option "kmsdev" "/dev/dri/card0" Option "Monitor-HDMI-2" Option "ZaphodHeads" "HDMI-2" EndSection Section "Monitor" Identifier "C Monitor" EndSection Section "Screen" Identifier "C" Device "C Card" Monitor "C Monitor" EndSection |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.