Mailing List Archive

Re: [syslinux] SYSLINUX 3.72 released
Dag Wieers wrote:
> On Thu, 25 Sep 2008, H. Peter Anvin wrote:
>
>> Well, it has to happen at some point, so I have officially pushed out
>> SYSLINUX 3.72. The big news in this release is of course the ISOLINUX
>> hybrid mode support (a single image which can be booted either from a
>> CD-ROM or from a hard disk/USB stick), but there are plenty of more
>> goodies in here, including the label listing (not technically
>> completion) on the Tab key.
>
> Features look very promising, however when compiling on RHEL4 and older, I
> get:
>
> gcc -Wp,-MT,setjmp.o,-MD,./.setjmp.o.d -Os -march=i386 -falign-functions=0 -falign-jumps=0 -falign-labels=0 -ffast-math -fomit-frame-pointer -std=gnu99 -m32 -g -mregparm=3 -DREGPARM=3 -D__COM32__ -nostdinc -iwithprefix include -I. -I./sys -I../include -W -Wall -Wpointer-arith -Wwrite-strings -Wstrict-prototypes -Winline -DDYNAMIC_CRC_TABLE -DPNG_NO_CONSOLE_IO -DPNG_NO_WRITE_SUPPORTED -DPNG_NO_MNG_FEATURES -DPNG_NO_READ_tIME -DPNG_NO_WRITE_tIME -D__ASSEMBLY__ -c -o setjmp.o setjmp.S
> setjmp.S:2:3: invalid preprocessing directive #arch
> setjmp.S:4:3: invalid preprocessing directive #setjmp
> setjmp.S:8:3: invalid preprocessing directive #The
> setjmp.S:9:3: invalid preprocessing directive #%
> setjmp.S:10:3: invalid preprocessing directive #%
> setjmp.S:11:3: invalid preprocessing directive #%
> setjmp.S:12:3: invalid preprocessing directive #%
> setjmp.S:13:3: invalid preprocessing directive #%
> setjmp.S:14:3: invalid preprocessing directive #<
> make[2]: *** [setjmp.o] Error 1
> make[2]: Leaving directory `/dar/build/syslinux-3.72/com32/lib'
> make[1]: *** [install] Error 2
> make[1]: Leaving directory `/dar/build/syslinux-3.72/com32'
> make: *** [install] Error 2
>
> Easy to fix ?
>

Yes, this is apparently due to cpp mistaking the comments at the top of
the file for preprocessor directives. Note that this code hasn't
changed since 2006, so I'm a bit confused why it pops up now; it might
be the Makefile changes (I have had a hard time to not get the specfile
to recompile too much.)

The biggest problem is that the code is quite frankly littered with
these, and changing them all would be a pain. I'm wondering if it is
the -std=gnu99 which is causing the trouble; if so there might be a sane
way to deal with it.

-hpa

_______________________________________________
SYSLINUX-announce mailing list
SYSLINUX-announce@zytor.com
http://zytor.com/mailman/listinfo/syslinux-announce
Re: [syslinux] SYSLINUX 3.72 released [ In reply to ]
Dag Wieers wrote:
>
> Looking at the code I have no clue how to make the for_each_pci_func()
> block into something acceptable. My C knowledge cannot even parse that
> construct :) I didn't know that C could do something like this.
>

C99 can, but not earlier versions. Anyway, I have checked in a pair of
patches to convert offending #-style comments to C-style comments in the
com32 part of the code; the rest of the code can stay as-is.

-hpa

_______________________________________________
SYSLINUX-announce mailing list
SYSLINUX-announce@zytor.com
http://zytor.com/mailman/listinfo/syslinux-announce
Re: [syslinux] SYSLINUX 3.72 released [ In reply to ]
Dag Wieers wrote:
>
> Patches work fine.
>
> This problem made it clear to me to keep the syslinux-test packages (based
> on pre-release) back in sync so we can find these kinds of problems before
> the actual release.
>
> I apologise for not having tested a 3.72-pre before the release. The signs
> were obvious I just ignored them :-/ Next time you advertise a pre-release
> as being close to a release I'll try them on RH9, RHEL3, RHEL4 and RHEL5
> again. Thanks again.
>

Thanks, I appreciate that. It would really cut down on this kind of
problems.

-hpa

_______________________________________________
SYSLINUX-announce mailing list
SYSLINUX-announce@zytor.com
http://zytor.com/mailman/listinfo/syslinux-announce