Le 30/08/2011 21:08, Wolfgang Denk a écrit :
Dear Albert ARIBAUD,
In message4E5D088B.7090702@aribaud.net you wrote:
I have just applied Wolfgang's updated ARM board removal patch set.
Currently, u-boot-arm/master builds 182 boards, of which:
- ELDK4.2 finds 37 boards with warnings or errors ( dnp1110 gcplus lart
I think we should try and separate the errors cases from the warnings only ones - those boards with build errors need to be fixed, or removed.
Right, but the current MAKEALL output does not make it easy to sort this out. Maybe we could change MAKEALL to distinguish between error and warning counts (and produce LOG/*.MAKELOG, LOG/*.ERR and LOG/*.WARN accordingly).
We can clean up warnings then, as we go on.
Agreed. However, I'd prefer it if no ARM board produced any warning -- granted, here it *looks like* many of these warnings are inocuous, and any new toolchain might produce some more, but if we don't keep the warning count at zero, then a less inocuous warning could creep in unnoticed.
- CodeSourcery 2010q1 finds 100 (!) boards with warnings or errors (
I guess it's additional warnings. You will see even more when using ELDK 5.x
(Cc:ing Rémy for USB stuff; for yaffs, I don't know whom I should Cc:)
yeaffs is generic code, i. e. it ends up on my plate.
There you are, then. :)
For both CS and ELDK, there are a lot of "arm-linux-ld: stubs.o: compiled for a big endian system and target is little endian arm-linux-ld: failed to merge target specific data of file stubs.o" which I'll look into.
You need a big endian ARM compiler for the ixp systems.
Got it -- actually, it is not a compiler issue, but a library issue: CS does compile for big-endian as well as little-endian, but does not provide big-endian libs (http://www.codesourcery.com/sgpp/lite/arm/portal/kbentry36, at least as I understand it). Apparently ELDK4.2 does not either. Would ELDK5.0 have them? I haven't gotten around to setting it up.
Meanwhile, I suggest that the current u-boot-arm/master be considered rc1-ready, with the effort until rc2 focused on finding and fixing code-wide bugs such as those two above.
Agreed.
If that is ok, I'll send out a pull request today.
Thanks a lot!
Done, with a few hour's delay.
Best regards,
Wolfgang Denk
Amicalement,