Re: panic in Linux emulation [was Re: 6.99.32: panic when starting X]

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

On Sun, Feb 23, 2014 at 12:07:33PM +0000, David Laight wrote:
> On Sun, Feb 23, 2014 at 12:32:05PM +0100, Thomas Klausner wrote:
> > On Sun, Feb 23, 2014 at 10:41:06AM +0100, Thomas Klausner wrote:
> > > After updating from 6.99.31/20140215 to 6.99.31/20140221, I started X
> > > and got a panic:
> > > 
> > > fatal protection fault in supervisor mode
> > > trap type 4 code 0 rip ffffffff808dbe2c cs 8 rflags 13286 cr2 
> > > ffff80023b989000 ilevel 0 rsp fffffe813ba2e9b0
> > > curlwp 0xfffffe813ba139c0 pid 328.1 lowest kstack 0xfffffe813ba2b280
> ...
> > > --- trap (number 4) ---
> > > usb_allocmem_flags() at netbsd:usb_allocmem_flags+0x6c
> > > ehci_allocm() at netbsd:ehci_allocm+0x2a
> ...
> > > After a reboot I tried again (while logged in via ssh, not via the USB
> > > keyboard like before) and it worked.
> > 
> > I could also start from the console, so it seems to have a random
> > element.
> > 
> > Why did I need to start so soon? Because of a second panic.
> > 
> > panic: kernel diagnostic assertion "len <= LINUX32_ELF_AUX_ENTRIES" failed: 
> > file "/archive/foreign/src/sys/compat/linux32/common/linux32_exec_elf32.c", 
> > line 244 
> > cpu7: Begin traceback...
> > vpanic() at netbsd:vpanic+0x1cd
> > kern_assert() at netbsd:kern_assert+0x5a
> > linux32_elf32_copyargs() at netbsd:linux32_elf32_copyargs+0x407
> ...
> > This happened during a bulk build, I guess building one of the suse
> > packages. I'll retry.
> 
> Hmmm... these could be 'random' panics.
> Might be worth looking for a latter working kernel.
> Or look through the commits for something suspicious and try the other
> side of it.
> 
> To even attempt to diagnose the crash we'd need a working kernel dump.
> (Do those work on amd64 at the moment - if you are willing to wait
> for the hours it takes to write out 16GB ...)

Oddly I also saw an X crash like that once, and then couldn't repeat it.
And I saw the linux problem yesterday when doing pkg_delete suse12_x11.
Must hunt for dumps...

Cheers,

Patrick