I have good news about this bug (can't resume: "Decompression yielded
xxx bytes instead of yyy.").
I recently tried to upgrade my machines (laptop + desktop) from
2.6.24.x/3.0-rc5 to tuxonice-2.6.26.git. Both machines are patched with
loop-aes and hibernate in a loop-aes-v3 encrypted partition.
This regression is still there on my laptop (filewriter), but I don't
have the problem on my desktop (swapwriter). So I tried swapwriter on
the laptop and it works! :)
Anyway, I discovered TOI can suspend on an unused swap partition. This
is even better than having a filewriter partition.
To sum up:
* I am very happy I am not stuck to 2.6.24.x/3.0-rc5.
* The regression only affects filewriter.
* I am not concerned anymore by this bug.
JM
Nigel Cunningham a écrit : > Hi.
>
> On Sun, 2008-05-11 at 19:11 +0200, Julien Muchembled wrote:
>> Nigel Cunningham a écrit :
>>> Do you get that bug on every attempt at resuming?
>> Yes.
>>
>>> When you try rc7, is it being applied to exactly the same kernel as rc5
>>> was being applied to?
>> Yes.
>> 16 days ago, I only tried to upgrade from 2.6.24.4/3.0-rc5 to
>> 2.6.25/3.0-rc7.
>>
>> I've just retried to upgrade TOI only (-> 2.6.24.4/3.0-rc7) and it fails
>> too.
>> (tried also 2.6.25.3 and 2.6.24.7 -> crash)
>
> Okay. Thanks.
>
>>> If you (temporarily) disable the encryption, does it work okay?
>> No! I am surprised to see that it also crashes without encryption.
>> More precisely:
>> /dev/sda1 / xfs
>> /dev/sda2 filewriter/lzf
>> /dev/loop0 ~jm xfs # not mounted
>
> Ok. That's helpful.
>
>> Attaching pictures for 2.6.25.3/3.0-rc7
>> (about ACPI errors: I have acpi.debug_level=0xF267A02F in cmdline)
>>
>> JM
>>
>> PS: Not sending to the ml due to the size of the photos
>
> Thanks!
>
> Nigel
(text/plain)