On Mon, 14 May 2012 22:58:58 +0200 Jes Sorensen <Jes.Sorensen@xxxxxxxxxx> wrote: > On 05/14/12 22:57, Paweł Sikora wrote: > > On Monday 14 of May 2012 22:30:32 Jes Sorensen wrote: > >> On 05/12/12 08:36, Paweł Sikora wrote: > >>> fyi, this release contains a hidden strict-aliasing-violation bug [1]. > >>> please fix this correctly or build mdadm with -fno-strict-aliasing option. > >>> > >>> [1] http://www.spinics.net/lists/raid/msg37298.html > >> > >> If you provide a patch, you're more likely to get the issue fixed than > >> by just telling other people to fix it for you. > > > > i have sent you a revert request with a proper fix 3 months ago > > and you have finally rejected it with "invalid commit message", > > so feel free to keep broken mdadm sources as long as you wish :> > > > > If you post a patch without including a proper commit message and > signed-off-by lines, which is standard practice, then that is equivalent > to you not having posted a patch in the first place. > > Don't blame others for not doing the basic work everybody else does when > posting patches. It really isn't hard to comply with the standard patch > posting rules! I decided a better approach is just to refresh from upstream: http://gcc.gnu.org/svn/gcc/branches/cilkplus/include/sha1.h http://gcc.gnu.org/svn/gcc/branches/cilkplus/libiberty/sha1.c have been copied into mdadm with just a tiny change to not include config.h Seems to work, and probably has the *right* fix. NeilBrown
Attachment:
signature.asc
Description: PGP signature
