Re: Solid lockup on latest benh 2.6 kernel + PB 12"
Hi Ben,
I've recompiled using the provided configs, and that nasty IN message is
gone.
However, the FB freeze still exists, even with video=ofonly.
Here's the dmesg attached:
However, I've narrowed the cause of this framebuffer hang (happens with
rivafb and ofonly video).
A long line of characters crash it. e.g. start at a prompt, hold down 'a'
key. By the time you get within about 3 chars of the right hand of the
screen, it locks up. As I said before, the machine is still responsive and
you can SSH in etc.
Actually, while ssh'd in, I spotted this appearing in dmesg:
Oops: kernel access of bad area, sig: 11 [#1]
NIP: C01AE394 LR: C01AE418 SP: CF36DCA0 REGS: cf36dbf0 TRAP: 0301
ot tainted
MSR: 00009032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
DAR: D10C0000, DSISR: 40000000
TASK = cfab30a0[306] 'bash' Last syscall: 4
GPR00: 00000000 CF36DCA0 CFAB30A0 00000300 00000010 000002F0 00000008
00000400
GPR08: 0000000F D10C0000 D10BFFE8 CFE5A1F8 280044E8
Call trace:
[c01a6cbc] fbcon_cursor+0x2b0/0x4ac
[c010f6b0] set_cursor+0xc0/0xe0
[c0113764] con_flush_chars+0x50/0x58
[c0113534] con_write+0x24/0x3c
[c0103e24] opost_block+0x150/0x258
[c0106014] write_chan+0x194/0x244
[c0100db8] tty_write+0x118/0x1f8
[c0059118] vfs_write+0xdc/0x128
[c0059218] sys_write+0x40/0x74
[c0007c6c] ret_from_syscall+0x0/0x4c
Think it might be relevant? :)
David
>
> On Thursday 19 February 2004 13:44, you wrote:
> > On Thu, 2004-02-19 at 04:47, David Pye wrote:
> > > Hi,
> > >
> > > my dmesg and kern.log seem full of info which floods the useful info
> > > you requested. I did spot it mentioning initialising OpenFirmware
> > > framebuffer driver. Any idea of how to get the whole dmesg output?
> > >
> > > Also, I get many copies of this error in kern.log
> > >
> > > IN from bad port 64 at c014f020
> >
> > That seem to indicate you have some legacy shit compiled in your
> > kernel. Are you using a pmac_defconfig ?
> >
> > > Regards the framebuffer, Interestingly, this does not lock up the fb:
> > >
> > > ls -1 -R /
> > >
> > > whereas
> > >
> > > ls -R / does. Column widths perhaps?
> > >
> > > append="video=ofonly" appears to have no effect on this lockup so far.
> > >
> > > Regards,
> > >
> > > David
> > >
> > > On Tuesday 17 February 2004 23:39, Benjamin Herrenschmidt wrote:
> > > > On Wed, 2004-02-18 at 09:45, David Pye wrote:
> > > > > I've rushed ahead and made the change to mine, and it doesn't seem
> > > > > to make any difference.
> > > > >
> > > > > Sometimes a hang on boot, but what DEFINITELY triggers it is doing
> > > > > something like ls /dev on a console. From an xterm it's fine, but
> > > > > from the console - bang the framebuffer locks solid. The box is
> > > > > still running fine though.
> > > >
> > > > Are you _certain_ you actually tested with video=ofonly ? That is
> > > > append=" video=ofonly" in yaboot.conf ? Send me a dmesg log of that
> > > > case so I can check if offb is really picking up the display.
> > > >
> > > > Ben.
Reply to: