Messages in this thread | | | Date | Mon, 20 Jul 2009 19:47:37 -0400 | From | Marc Dionne <> | Subject | Re: Linux 2.6.27.27 |
| |
On 07/20/2009 06:08 PM, Linus Torvalds wrote: > > On Mon, 20 Jul 2009, Krzysztof Oledzki wrote: >> No problem. Please let me know what should I do to help tracking this issue. > > Can you build two kernels: one with -fwrapv, and one with > -fno-strict-overflow, and then verify that > > - they are otherwise identical (ie exact same source code, same compiler > etc) > > - verify that yes, the -fwrapv kernel works, the other does not. Just to > avoid the confusion that obviously exists with Debian/sid binutils > upgrades that _also_ happens result in nonbootable kernels. > > - upload the 'vmlinux' images somewhere (I'm not sure what the limits for > binary attachments are at the kernel bugzilla, but that would be the > logical place) > > In fact, it would be nice to have a third "identical" kernel build, except > with neither -fwrapv/-fno-strict-overflow. > > Linus
I might be seeing a slightly different bug, but in case it's helpful, the behaviour here on Fedora rawhide with gcc-4.4.0-14.x86_64 and binutils-2.19.51.0.11-27.fc12.x86_64 is that I get various .o files that come out as completely empty files (or in one case as a precisely 64K sized file that gives a "File format not recognized" error"), and the latest 2.6.31-rc git can't be built at all.
If I replace -fno-strict-overflow with -fwrapv in Makefile everything builds and runs fine.
Interestingly though, "-fno-strict-overflow -v" also gives me a good kernel, and comparing the assembly for one of the affected files doesn't show any difference between -fwrapv and -fno-strict-overflow.
Marc
| |