Hi, Let me elaborate problem in detail. For buffered IO data are copied into memory pages. For this case, the write IO is not submitted (generally). In the background opportunistic cleaning of dirty pages takes place and IO is generated to the device. An IO error is observed on this path and application is not informed about this. Summarizing flushing of dirty page fails. And probably, this page is dropped but in fact it should not be. So if above situation happens between application write and sync then no error is reported. In addition after some time, when the application reads the same LBA on which IO error occurred, old data content is fetched. We did own fault injector in order to do error in specific condition described above. Regards, Mariusz. -----Original Message----- From: Andreas Dilger [mailto:adilger@xxxxxxxxx] Sent: Thursday, June 2, 2016 21:32 To: Barczak, Mariusz <mariusz.barczak@xxxxxxxxx> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>; Jens Axboe <axboe@xxxxxxxxx>; Alexander Viro <viro@xxxxxxxxxxxxxxxxxx>; linux-mm@xxxxxxxxx; linux-block@xxxxxxxxxxxxxxx; linux-fsdevel@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Wysoczanski, Michal <michal.wysoczanski@xxxxxxxxx>; Baldyga, Robert <robert.baldyga@xxxxxxxxx>; Roman, Agnieszka <agnieszka.roman@xxxxxxxxx> Subject: Re: [BUG] Possible silent data corruption in filesystems/page cache On Jun 1, 2016, at 3:51 AM, Barczak, Mariusz <mariusz.barczak@xxxxxxxxx> wrote: > > We run data validation test for buffered workload on filesystems: > ext3, ext4, and XFS. > In context of flushing page cache block device driver returned IO error. > After dropping page cache our validation tool reported data corruption. Hi Mariusz, it isn't clear what you expect to happen here? If there is an IO error then the data is not written to disk and cannot be correct when read. The expected behaviour is the IO error will either be returned immediately at write() time (this used to be more common with older filesystems), or it will be returned when calling sync() on the file to flush cached data to disk. > We provided a simple patch in order to inject IO error in device mapper. > We run test to verify md5sum of file during IO error. > Test shows checksum mismatch. > > Attachments: > 0001-drivers-md-dm-add-error-injection.patch - device mapper patch There is already the dm-flakey module that allows injecting errors into the IO path. Cheers, Andreas -------------------------------------------------------------------- Intel Technology Poland sp. z o.o. ul. Slowackiego 173 | 80-298 Gdansk | Sad Rejonowy Gdansk Polnoc | VII Wydzial Gospodarczy Krajowego Rejestru Sadowego - KRS 101882 | NIP 957-07-52-316 | Kapital zakladowy 200.000 PLN. Ta wiadomosc wraz z zalacznikami jest przeznaczona dla okreslonego adresata i moze zawierac informacje poufne. W razie przypadkowego otrzymania tej wiadomosci, prosimy o powiadomienie nadawcy oraz trwale jej usuniecie; jakiekolwiek przegladanie lub rozpowszechnianie jest zabronione. This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). If you are not the intended recipient, please contact the sender and delete all copies; any review or distribution by others is strictly prohibited. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href