On Monday, September 26, 2011 06:10:37 PM Mike Frysinger wrote: > On Monday, September 26, 2011 05:03:46 Marek Vasut wrote: > > On Monday, September 26, 2011 09:25:36 AM Wolfgang Denk wrote: > > > Mike Frysinger wrote: > > > > > The assigned variable can be removed because the pointers are > > > > > volatile so accesses to their addresses are always generated. > > > > > > > > i think the stores to a var were added to avoid a gcc warning, but if > > > > newer versions don't warn anymore, should be fine. > > > > > > Obviously we have to check if older tools still accept this. > > > > Hi, > > > > as I was never completely sure about this file from the begining, I'm all > > for it. > > > > Btw. who's supposed to merge the rest of the series anyway ? > > > > > Marek, which tool chains did you use for testing? > > > > Hand-built gcc4.6.0+debian patches and gcc4.4.6+debian patches: > > > > arm-linux-gnueabi-gcc-4.4 (Debian 4.4.6-7) 4.4.6 > > arm-linux-gnueabi-gcc-4.6 (Debian 4.6.0-14) 4.6.1 20110616 (prerelease) > > > > There was someone on the irc who (unsuccessfully) tried to compile uboot > > with gcc 2.xx recently. Maybe we should impose some lower bound for > > compiler version, though I guess 4.4 is too high. Suggestions ? > > i think gcc-3.x has been broken for a while but no one has noticed
I guess we should go the kernel way -- make it 4.2 and be done with it. > -mike _______________________________________________ U-Boot mailing list [email protected] http://lists.denx.de/mailman/listinfo/u-boot
