I've been using the pre-FC2 Raw Hide and FC2 kernels (including 2.6.5-1.358 2.6.5-1.351 2.6.5-1.346 2.6.5-1.339 2.6.5-1.332 and a lange number of earlier ones) recompiled with CONFIG_SOFTWARE_SUSPEND enabled without much problems. However when I try recompiling 2.6.6-1.427 with CONFIG_SOFTWARE_SUSPEND, I get: ... LD .tmp_vmlinux1 arch/i386/power/built-in.o(.text+0x1a3): In function `do_magic': : undefined reference to `swsusp_pg_dir' make: *** [.tmp_vmlinux1] Exit 1 grep suggests that the only occurrence of swsusp_pg_dir in the sources are: ./arch/i386/power/pmdisk.S: movl $swsusp_pg_dir-__PAGE_OFFSET,%ecx ./arch/i386/power/swsusp.S: movl $swsusp_pg_dir-__PAGE_OFFSET,%ecx According to linux.bkbits.net, the "swsusp_pg_dir" first appeared in swsusp.S and pmdisk.S as a result of the following patch: http://linux.bkbits.net:8080/linux-2.5/cset@40b3691eO9MJaSUywupm4EydmK80Iw But for some reason that patch is present in 2.6.6-1.427 only partially - the arch/i386/mm/init.c part is missing!
*** Bug 125986 has been marked as a duplicate of this bug. ***
Still there with 2.6.6-1.435
Created attachment 101369 [details] A patch to arch/i386/mm/init.c that fixes this This is a patch based on http://linux.bkbits.net:8080/linux-2.5/cset@40b3691eO9MJaSUywupm4EydmK80Iw
Me too with 435.2.1 It also happens with Planet CCRMA 2.6.7-1.437.1.11.fc2.ccrma
Still happens with 2.6.7-1.492 as well.
Tested 2.6.7-1.492 with attachment (id=101369) and it does build the rpm.
Still there with 2.6.8-1.521
Still there with kernel-2.6.9-1.640 in FC3 test. Patch (id=101369) still applies cleanly and appears to get the build to work. (Haven't booted it yet though.)
Still there with 2.6.9-1.667 in FC3 final release, but I haven't tried the patch yet.
*** This bug has been marked as a duplicate of 126342 ***
I disagree with this being duped against a generic "custom kernels" bug. This bug is about: 1) Red Hat/Fedora introducing a problem that does not exist in kernel.org kernels 2) A problem for which a fix is very simple and a patch is attached to this bug. 3) A problem that affects quite a number of people (there are 4 "mee too"s and one dup), that is readily reproducible and that existed for quite a few version of Red Hat/Fedora - provided kernels. P.S. It apperars that the linux-2.6.0-4g4g.patch is the one responsible for this problem.
too bad. its not something we support, hence not something we'll spend time on fixing. *** This bug has been marked as a duplicate of 126342 ***
> its not something we support, hence not something we'll spend time on > fixing. I am not asking you too - just asking to include a trivial patch that's already attached here...
Created attachment 107451 [details] Corrected linux-2.6.0-4g4g.patch This is the corrected version of the linux-2.6.0-4g4g.patch. I took the linux-2.6.0-4g4g.patch from 2.6.9-1.681_FC3 sources and updated the arch/i386/mm/init.c portion of it.
Created attachment 107452 [details] Difference between the original linux-2.6.0-4g4g.patch and the corrected one Note that this is a "level 2 diff" (i.e. a diff between two diffs). It shows the differences between the linux-2.6.0-4g4g.patch in 2.6.9-1.681_FC3 sources and the one in attachment 107451 [details]. P.S. Who is the maintainer of linux-2.6.0-4g4g.patch? Is [s]he aware of this bug?
As far as I can tell, this is finally fixed in 2.6.9-1.724_FC3.