Bug#1076372: Re.: linux-image-6.5.0-0.deb12.4-amd64: ext4 file corruption with newer kernels
- To: Stefan <debian@simg.de>, 1076372@bugs.debian.org
- Subject: Bug#1076372: Re.: linux-image-6.5.0-0.deb12.4-amd64: ext4 file corruption with newer kernels
- From: Salvatore Bonaccorso <carnil@debian.org>
- Date: Wed, 25 Sep 2024 20:57:40 +0200
- Message-id: <[🔎] ZvRdJOMpdiSpCeex@eldamar.lan>
- Reply-to: Salvatore Bonaccorso <carnil@debian.org>, 1076372@bugs.debian.org
- In-reply-to: <64a82ea9-6b85-4863-b44a-ff71e5e9d4f8@simg.de>
- References: <172104075060.7102.3621600478475051128.reportbug@ws7> <71c73e72-e505-477c-b199-fdc351773d2a@simg.de> <172104075060.7102.3621600478475051128.reportbug@ws7> <ZqQQpaDyhOJr2QJ2@eldamar.lan> <172104075060.7102.3621600478475051128.reportbug@ws7> <bf152ceb-9e44-44e1-be5d-d5a12f5b22ab@simg.de> <172104075060.7102.3621600478475051128.reportbug@ws7> <ZrugTv7d5j8sZqzq@eldamar.lan> <172104075060.7102.3621600478475051128.reportbug@ws7> <64a82ea9-6b85-4863-b44a-ff71e5e9d4f8@simg.de> <172104075060.7102.3621600478475051128.reportbug@ws7>
Hi Stefan,
On Wed, Aug 14, 2024 at 01:22:05AM +0200, Stefan wrote:
> Hi Salvatore,
>
> sorry, I had not the time to run all tests I planned.
>
> Here is what I found out so far:
>
> * The errors can be reproduced with the program `f3`, see
> https://fight-flash-fraud.readthedocs.io/en/latest/index.html. Errors
> are reported as "overwritten sectors", which probably means that they
> are address errors. The strange thing is, that the errors seem to
> disappear after the program runs a while (in contrast to the older test;
> log file enclosed). I tried to induce the/more errors (`stress-ng
> --cpu`, deleting files while f3 is running), but had no success. (It may
> be an issue with the pattern generator of `f3`.)
>
> * The errors cannot be reproduced on other systems (tested 3 ones), so
> it is hardware specific.
>
> * The bug is not fixed in 6.10 kernels
>
> I had not sufficient time to test another nvm. This would reveal whether
> it is a problem with the mainboard+CPU or with the nvm.
Did you had a chance to do these further tests?
In meanwhile we have retitled the bug to make it more specific to the
Longsys/Lexar NM790 NVMe drive.
Regards,
Salvatore
Reply to: