On Tue, Mar 18, 2014 at 12:20:18AM +0100, Julian Andres Klode wrote: > Package: src:linux > Version: 3.13.5-1 > Severity: normal > > 3.13.5 introduces a regression in the xhci code that causes mass > storage to not work correctly. In my case, the experience was that > first writes failed misteriously, the filesystem was corrupted > and then it could not be mounted anymore and I saw I/O errors.
It might even cause data loss. If the disks can get mounted and we start a write, some things could get lost or the disk corrupted, given that the FAT on my USB stick was corrupted after writing to it. So the bug should probably be RC. > > The problem seems to be known upstream > > https://lkml.org/lkml/2014/3/6/284 > > and fixed in 3.14-rc6. I hope the relevant commits (reverts) > will be backported to 3.13.7. If a stable release includes: > Revert "xhci 1.0: Limit arbitrarily-aligned scatter gather." > this most likely means that this bug is fixed. As far as I can tell, the needed revert commits are precisely: e2ed511400d41e0d136089d5a55ceab57c6a2426 Revert "xhci 1.0: Limit arbitrarily-aligned scatter gather." 469d417b68958a064c09e7875646c97c6e783dfc Revert "USBNET: ax88179_178a: enable tso if usb host supports sg dma" -- Julian Andres Klode - Debian Developer, Ubuntu Member See http://wiki.debian.org/JulianAndresKlode and http://jak-linux.org/. Please do not top-post if possible. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
