On Thu, 13 Oct 2011 12:48:54 -0500 Steve French <smfrench@xxxxxxxxx> wrote: > Any sign of reconnects in the failure case? (/proc/fs/cifs/Stats for > example) > > Can you add (or flip a debug message from cFYI to cERROR) a debug message > for any write errors > in SMB write ...? > Also, what mount options are you using? > On Thu, Oct 13, 2011 at 12:36 PM, Pavel Shilovsky <piastryyy@xxxxxxxxx>wrote: > > > 2011/10/13 Jeff Layton <jlayton@xxxxxxxxxx>: > > > On Thu, 13 Oct 2011 16:46:51 +0400 > > > Pavel Shilovsky <piastryyy@xxxxxxxxx> wrote: > > > > > >> 2011/10/13 Jeff Layton <jlayton@xxxxxxxxxx>: > > >> > On Thu, 13 Oct 2011 11:28:29 +0400 > > >> > Pavel Shilovsky <piastryyy@xxxxxxxxx> wrote: > > >> > > > >> >> I tested your current cifs-3.2 branch and faced with strange results > > >> >> (assuming Windows 7 share is mounted to /mnt/share): > > >> >> > > >> > > > >> > I've not seen this, but I'll try to reproduce it in a little while. > > >> > > > >> >> piastry@adventure:~$ dd if=test1.iso of=/mnt/share/test.iso > > bs=100000 > > >> >> > > >> > > > >> > Ok, so test1.iso is in your home directory and you're dd'ing it > > >> > to /mnt/share/test.iso (which is a cifs mount) > > >> > > > >> >> piastry@adventure:~$ dd of=test.iso if=/mnt/share/test.iso bs=100000 > > >> >> > > >> > > > >> > Then you copy test.iso back from the share to your homedir > > >> > > > >> >> piastry@adventure:~$ md5sum test.iso > > >> >> 5eee1675f46ac594dd7800c37e3ba944 test.iso > > >> >> > > >> > > > >> > Checksum of test.iso in your homedir and end up with the above... > > >> > > > >> >> piastry@adventure:~$ md5sum /mnt/share/test.iso > > >> >> 5eee1675f46ac594dd7800c37e3ba944 /mnt/HOME-share/test.iso > > >> >> > > >> > > > >> > The checksum matches the (apparent) one on the share > > >> > > > >> >> piastry@adventure:~$ md5sum test1.iso > > >> >> 10d672583cab035e2b55048fbfbbf7ca test1.iso > > >> >> > > >> > > > >> > But does not match the original. > > >> > > > >> >> Then, I umounted and mounted it again with 3.1-rc9 cifs module - > > >> >> everything was fine. Then, I load previous "buggy" one and everything > > >> >> was fine too. > > >> >> > > >> >> So, md5sums are not equal. Any ideas? > > >> >> > > >> > > > >> > Not something I've seen in my testing. Some questions: > > >> > > > >> > 1) where is your home directory? is it on a local disk? > > >> > > >> It's on ext3 local hard drive. > > >> > > >> > > > >> > 2) when you say "everything was fine" what checksum did you get? > > >> > > >> all checksums matched the original. > > >> > > >> > > > >> > 3) if this test is repeatable, is there any discrepancy in the size of > > >> > the file? IOW, when you do the md5sum, it would be good to also do a > > >> > "stat" against the file the i_size is coming up correctly after each > > >> > step. > > >> > > >> The size was the same - 415 MB. One time it happened I calculated the > > >> number of different characters in these files - it was 127 (I don't > > >> know how it can help, but anyway). > > >> > > > > > > I tried this too, but haven't been able to reproduce it. Grabbed a > > > random .iso file which is a bit bigger than yours -- 615M, and tried > > > several of these tests and the checksum always came out the same... > > > > > > Do any of the dd commands return an error? > > > > > > > No. It's hard to catch - I faced with it only twice while testing it > > (for about 30 runs). > > > > -- > > Best regards, > > Pavel Shilovsky. > > -- > > To unsubscribe from this list: send the line "unsubscribe linux-cifs" in > > the body of a message to majordomo@xxxxxxxxxxxxxxx > > More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > > > -- > Thanks, > > Steve -- Jeff Layton <jlayton@xxxxxxxxxx> -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
