This is great news. Thank you. s3nt fr0m a $martph0ne, excuse typ0s On Aug 29, 2014 1:24 PM, "Anton Ivanov" <anton.iva...@kot-begemot.co.uk> wrote:
> On 09/03/14 21:35, Mattia Dongili wrote: > >> On Sat, Mar 08, 2014 at 07:04:56AM +0000, Anton Ivanov wrote: >> >>> Package: user-mode-linux >>> Version: 3.2-2um-1+deb7u2+b1 >>> Severity: grave >>> Tags: patch >>> Justification: causes non-serious data loss >>> >>> Dear Maintainer, >>> >>> This bug is perennial. If we go through old bugs with >>> "cannot reproduce" tag 50% of them are this one, the other >>> 50% are the "you should not use pipe for interprocess IPC" >>> which we will submit shortly. >>> >>> Manifestation of the problem - UML dies on startup for no >>> reason with a memory corruption message. Occurs only on >>> heavily loaded systems and usually when running a lot of >>> UMLs. >>> >> Thanks for the patch. >> I have noticed that you submitted these patch-set (together with the >> other two you sent here and more) upstream and they will be in the >> stable branch. >> The easiest path here is also to go through the stable release of >> linux-source where uml is built from. I'll keep an eye on the stable >> tree but it'd be very helpful if you could add the stable tree commit >> ids once the patches get included. Same story for the other two bugs. >> > > All 3 bugs have now patches submitted upstream. I have submitted our other > improvements as well. > > While they do not make a speed daemon of uml userspace they get it > reasonably close to kvm. Kernel itself is now faster than qemu-kvm for most > networking stuff. > > A. > > >> Thanks! >> > >
