On 12/01/2013 01:00 PM, Hans-Christian Egtvedt wrote:
Around Sun 01 Dec 2013 12:26:08 -0800 or thereabout, Greg Kroah-Hartman wrote:
On Sun, Dec 01, 2013 at 12:19:04PM -0800, Guenter Roeck wrote:
Hi Greg,
the current -stable queue for 3.10 results in a build failure for avr32:atngw100mkii_evklcd101_defconfig.
lib/lib.a(vsprintf.o): In function `pointer':
vsprintf.c:(.text+0x17dc): relocation truncated to fit: R_AVR32_16N_PCREL against symbol `_ctype' defined in .text section in lib/lib.a(ctype.o)
make: *** [vmlinux] Error 1
Introduced by commit 312b4e22695 ("vsprintf: check real user/group id for %pK").
Reverting that commit fixes the problem. Not really sure ... no idea, really ...
what to do about it. Later kernel versions seem ok, so maybe there is another
patch which needs to be back-ported. Copying avr32 maintainers and Geert for feedback.
That's really odd, thanks for letting me know.
I'll keep the patch in the tree for now, until one of the avr32 people
can tell me what to do.
Stuff in lib/ is C, right? So the compiler/linker should pick appropriate
jump instruction, weird.
AVR32 arch got some fixes, but I would assume they are not related to this
7a2a74f4b856993218aa7cdeeb6c3103101340db avr32: setup crt for early panic()
and
d617b338bbfdd77e9cbd8e7dc949cee3dd73d575 avr32: fix out-of-range jump in
large kernels
I would be surprised if they are related to the code in lib/.
No, doesn't help (nor apply cleanly).
Turns out that applying patch 9196436 (vsprintf: ignore %n again) fixes the problem.
Maybe some jump distance is really close to the architecture limit,
or there is some side effect of the location of calls to _tolower().
Question is if 9196436 can or should be applied ... it does fix a possible
security vulnerability, so maybe.
Guenter
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html