-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Jason Wessel wrote: > Tom Rini wrote: >> On Fri, May 04, 2007 at 11:02:55AM -0500, Jason Wessel wrote: >> >>> Pete/Piet Delaney wrote: >>> >>>> -----BEGIN PGP SIGNED MESSAGE----- >>>> Hash: SHA1 >>>> >>>> gcc 4.1.1 is still a supported compiler by kernel.org; since we have >>>> what appears to be reasonable fix and it will prevent folks who happen >>>> to be using 4.1.1 from having weird problems it seems reasonable to >>>> apply the fix. >>>> >>>> >> That's the other thing. I kind of think that in my git tree I reworked >> things so that it simply wasn't a problem, possibly by just getting rid >> of all of the weak functions, and the number of them that were needed. >> I think it was the case that it was always: >> - The function used was the weak one. >> - The non-weak version was an "x86'ism" and the same on i386/x86_64/ia64 >> for just about everything. >> >> > > Given that it is all working at this point on the most common archs, > with gcc that you can obtain publicly, I am inclined not to kludge for > any particular specific compiler. Thanks for work on the cleanup Tom. > I would not recommend any further work around the "weak symbol" issue > unless there is a further known problem.
Last evening I asked Andrew about it and his option is that the kernel supports being compiled with gcc 4.1.1. So that means it likely doesn't currently have the weak decl problem. The kernel has only a couple dozen (weak) declarations and interesting has a __weak decl in include/linux/compiler-gcc.h and a note about avoiding a gcc bug for ppc64. Looks like __weak isn't used. Do you really want to be rejected from mainline because you break the kernel for some gcc compilers? Your "kind of think" isn't extremely comforting. How about having Milind Dumbare double check your reworking of things, perhaps with a gcc 4.1.1 compiler to verify that we won't in fact be adding a bug to the kernel. I've never even felt it was ever necessary or best to use the weak paradigm; we likely could code this in a more conventional form. Anyway, I really wonder if it might be best for Milind should double check your reworking. I was going to pull out a CVS snapshot of your branch and see if I see a problem but it sounds like it might only be in your git repository. I tried looking thru my email but didn't find the original discussion of the problem. As I vaguely recall if both decl occurred it ended up compiling in the default instead of the architecture specific version. Milind's change of moving the decls to kgdbarchlib.c seems easy. If you don't have a kernel.org account yet we could put a copy of your git repository there in my directory for now so folks, especially Milind, can clone it; or perhaps you have a server available where you can publish it for us to clone and checkout. If you send me tar ball I'll gladly put it on kernel.org to facilitate integration. - -piet > > Jason. > -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFGO+2hJICwm/rv3hoRArcsAKCD/Ac14uGyoMkZCzMuzrv9zRa9yQCeM2IG NjPBoCnLWr6TWlzw7Ob+nEc= =MMZd -----END PGP SIGNATURE----- ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Kgdb-bugreport mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/kgdb-bugreport
