Re: "EXT3-fs error" after resume from s2ram

From: Christof Warlich
Date: Tue Jul 07 2009 - 13:42:19 EST


Robert Hancock schrieb:
On 07/07/2009 03:13 AM, Christof Warlich wrote:
Christof Warlich schrieb:
Do you think I can fix the issue and circumvent the BIOS by passing
the real geomerty of the drive as kernel boot parameters? This may
also have the nice side effect that I could utilize the full capacity
of the drive. I'll definitly do some experiments in this direction and
let you know immediately if I succeed. Please let me know if this may
be a waste of time or if you know an appropriate fix.
I retried with the boot parameter hda=16383,16,63 according to the
values that I got from fdisk, but without success. Anyway, I would have
expected that the kernel may have had a way to set the LBA48 geometry
instead of CHS, but it looks like I'm on the wrong track here...

That's an old IDE parameter, not a libata one, and 160GB is way too large to be settable using CHS. What you likely want to do is pass the ignore_hpa=1 option to libata when it's loaded - either libata.ignore_hpa=1 on kernel command line, or in /etc/modprobe.conf if it's a module (how exactly to do this depends on your distribution). That way the protected area on the disk will get disabled and you can use the full capacity.
Hmm - that doen't seem to fix it :-(:

After the reboot, fdisk still reports 137.4GB, and dmesg now contains the line:

ata1.00: device aborted resize (268435456 -> 312581808), skipping HPA handling

instead of

ata1.00: HPA detected: current 268435456, native 312581808

that we have seen without the libata.ignore_hpa=1 kernel command line option.

After s2ram -f -p, I still see the EXT3-fs error on resume. Is it worth to take another log from this suspend /resume cycle? begin:vcard
fn:Christof Warlich
n:Warlich;Christof
org:Warlich Consulting
adr:;;Am Haselhang 7;Hilpoltstein;Bayern;91161;Deutschland
email;internet:christof@xxxxxxxxxxxx
tel;work:09174/492914
tel;home:09174/492914
x-mozilla-html:TRUE
url:www.warlich.name
version:2.1
end:vcard