On 12/15/05, Christian Sühs chris@suehsi.de wrote:
BTW: GCC 2.96 compiles the flashrom tool without any warnings and errors. At time it seems the tool works correct. It recognized the right chipset and the right EEProm Chip.
What flash part (EEProm) is on that board ?
But unfortunatly gcc 2.96 fails to compile the LinuxBios Image :D However, if I recieve my chip I will try to flash LB with the 2.96 compiled flashtool and the gcc 3.3 compiled image ;) and see what happens.
What SVN rev are you using? I thought Stefan reciently fixed up all the compile and alignment issues to work with newer compilers.
Its ok if flashrom and the linuxbios image are compiled with different compilers.
-- Richard A. Smith
Richard Smith schrieb:
On 12/15/05, Christian Sühs chris@suehsi.de wrote:
What flash part (EEProm) is on that board ?
It is a SST39SF020A @ 0xFFFC0000 the chipset is VIA VT8235
The epia ML is a downgraded board like the ME, but without FDC, 1394, LVDS and few things more :(
What SVN rev are you using? I thought Stefan reciently fixed up all the compile and alignment issues to work with newer compilers.
Yeah, i have read this. I have tried out two versions of LB, first a two days old snapshot and second the 2100 branch with the same results. The changes by Stefan should be inside the first
Its ok if flashrom and the linuxbios image are compiled with different compilers.
Tomorrow I receive my Chip, then I will start :D and sorry that I haven't CC all messages to the mailinglist. Sometimes I'm a stupid guy ;)
* Richard Smith smithbone@gmail.com [051215 17:25]:
On 12/15/05, Christian Sühs chris@suehsi.de wrote:
BTW: GCC 2.96 compiles the flashrom tool without any warnings and errors. At time it seems the tool works correct. It recognized the right chipset and the right EEProm Chip.
weird. gcc 3.3.5 from SUSE 9.3 and gcc 4.0.2 from SUSE 10 work fine. You should definitely avoid using gcc 2.96 when possible.
What SVN rev are you using? I thought Stefan reciently fixed up all the compile and alignment issues to work with newer compilers.
I did indeed. Just tested r2145 with gcc version 4.0.2 20050901 (prerelease) (SUSE Linux)
I'll look into those warnings asap though I am a bit confused. Usually gcc 4.x is a lot more verbose with warnings than 3.3.x
Stefan
weird. gcc 3.3.5 from SUSE 9.3 and gcc 4.0.2 from SUSE 10 work fine. You should definitely avoid using gcc 2.96 when possible.
Unfortanetly, with gcc 2.96 I'm not able to compile the LB image itself :D
Furthermore, I work on a selfmade minimal distro, but I will try to compile a working gcc version to compile both, flashtool and the image without any warnings
Currently I'm wondering about the differences in gcc. For flashtool I get warnings for comparisons between signed and unsigned only, without the -Werror Flag and the gcc 3.3 version. Why isn't gcc backwards compatible?
Now I have the tool compiled with gcc 2.96 and it works, but compiling the image with gcc 3.3 results in the same warnings and a few warnings more. It could be, that the image won't work :(
However, I will test the results the next days and stay tuned :D
regards chris
@Stefan
I think i can mail to in german, is it so ??
* Christian Sühs chris@suehsi.de [051215 22:22]:
Unfortanetly, with gcc 2.96 I'm not able to compile the LB image itself :D
As I said: Don't use 2.96, except the world goes under otherwise. ;)
Furthermore, I work on a selfmade minimal distro, but I will try to compile a working gcc version to compile both, flashtool and the image without any warnings
Ah, maybe crosstool is of interest for you. It allows you to build different combinations of gcc, binutils and libc very easily. Just google for it.
Currently I'm wondering about the differences in gcc. For flashtool I get warnings for comparisons between signed and unsigned only, without the -Werror Flag and the gcc 3.3 version. Why isn't gcc backwards compatible?
In this case removing -Werror should be fine. A gdb run of the segmentation fault would be very interesting. I don't have an 8.2 machine here anymore, unfortunately.
gcc has never been without such side effects, and using stuff like -Werror might give safety on a couple of systems, but also introduces compilation problems on others. Is your gcc 3.3 a prerelease? (ie. as shipped with SUSE a couple of times) those prereleases often handle code better than the final releases, but they might have warnings enabled that never occur again in any later version.
Now I have the tool compiled with gcc 2.96 and it works, but compiling the image with gcc 3.3 results in the same warnings and a few warnings more. It could be, that the image won't work :(
Does the segfault you saw happen with the 2.96 binary as well? What did you do to get around it?
@Stefan I think i can mail to in german, is it so ??
Indeed. But please make sure to leave the mailing list off the senders list in german mailings. ;-)
Regards, Stefan
-
Christian Sühs -
Richard Smith -
Stefan Reinauer