On Mon, 2013-04-01 at 18:37 +0000, Thorsten Glaser wrote:
> Hi Ben,
>
> the latest upload FTBFS on m68k:
>
> CC [M] drivers/hid/hid-microsoft.o
> /tmp/buildd/linux-3.2.41/drivers/hid/hid-microsoft.c: In function 'ms_report_fixup':
> /tmp/buildd/linux-3.2.41/drivers/hid/hid-microsoft.c:52:18: error: macro "memcmp" passed 6 arguments, but takes just 3
> /tmp/buildd/linux-3.2.41/drivers/hid/hid-microsoft.c:51:5: error: 'memcmp' undeclared (first use in this function)
> /tmp/buildd/linux-3.2.41/drivers/hid/hid-microsoft.c:51:5: note: each undeclared identifier is reported only once for each function it appears in
> make[7]: *** [drivers/hid/hid-microsoft.o] Error 1
>
> 49 /* the same as above (s/usage/physical/) */
> 50 if ((quirks & MS_RDESC_3K) && *rsize == 106 &&
> 51 !memcmp((char []){ 0x19, 0x00, 0x29, 0xff },
> 52 &rdesc[94], 4)) {
> 53 rdesc[94] = 0x35;
> 54 rdesc[96] = 0x45;
> 55 }
>
> I asked waldi in IRC whether this was known, and he said it
> built on other arches, so maybe m68k is the only one to use
> a macro; he suggested to put it upstream, and, well, upstream
> for Linux 3.2 stable series is you ;-)
This is fugly code, but m68k's string library is weird and needs fixing.
Geert found another example of that recently:
<http://article.gmane.org/gmane.linux.kernel.next/26701>.
Anyway: the basic rule for stable bug fixes is that the bug must be
fixed (or never introduced) in mainline first. So check mainline first.
Ben.
> I’ll let Geert chime in, I think the issue was discussed on
> one of the m68k lists already, but we’ll need it fixed in
> stable (and Debian) eventually.
>
> Thanks,
> //mirabilos
--
Ben Hutchings
DNRC Motto: I can please only one person per day.
Today is not your day. Tomorrow isn't looking good either.
Attachment:
signature.asc
Description: This is a digitally signed message part