Control: tag -1 upstream confirmed
On Wed, 2014-11-05 at 15:02 +0100, Evgeni Golov wrote:
Package: src:linux Version: 3.16.7-1 Severity: normal
Hi,
while testing the new Grml release, I noticed an interesting hang of QEMU when the machine is booted. The initial testing was done using [1] in QEMU from Sid (2.1+dfsg-5), but I also could reproduce it with the same QEMU and a fresh Jessie install with both, 3.16.5-1 and 3.16.7-1.
To reproduce:
- boot a fresh Jessie with 3.16 kernel in QEMU and vga=791 as bootparam
- modprobe cirrus modeset=1
- machine hangs
I am not sure this will happen on real cirrus HW, as I don't have any, but QEMU is quite widespread and cirrus is the default VGA driver.
[...]
The cirrus KMS driver only binds to QEMU Cirrus devices, not real hardware. It has been working for me, but I can reproduce the apparent hang with the 'vga=791' parameter.
It's not actually hanging, though - using a serial console, I see the error messages:
[ 1.976874] [drm:cirrus_vram_init] *ERROR* can't reserve VRAM [ 1.978067] cirrus 0000:00:02.0: Fatal error during GPU init: -6
and I can log in there. But, as the driver detects this error after the point of no return (remove_conflicting_framebuffers()), the display remains broken.
Ben.