Philip Guenther: > > This happens at least on amd64 and sparc64, but not on i386: > > -------> > > cc -std=gnu99 -O2 -pipe -Wall -W -o .libs/tiffmedian tiffmedian.o > > -L/usr/local/lib -L../libtiff/.libs -ltiff ../port/.libs/libport.a -ljpeg > > -lz -lm -Wl,-rpath,/usr/local/lib > > /usr/bin/ld: ../libtiff/.libs/libtiff.so.38.2: invalid string offset > > 3503345872 >= 4739 for section `.dynstr' > > Offset beyond end of section? Looks like libtiff.so.38.2 is corrupt > to me. I would remove all its objects and rebuild it and see if the > problem persists,
The problem is reproducible on two archs. I don't see where corruption could occur. ... It doesn't happen on alpha. ... It doesn't happen with jpeg6. Interesting. > > This popped up on i386: > > -------> > > cc -O2 -pipe -g -o .libs/epeg epeg_main.o -L/usr/local/lib > > -L../../src/lib/.libs -lepeg -ljpeg -Wl,-rpath,/usr/local/lib > > ../../src/lib/.libs/libepeg.so.0.0: undefined reference to `no symbol' > > Weird: if I'm reading the ld source correctly, the code to look up a > symbol by name was passed a NULL pointer. Again, a corrupt library > seems likely. This was the only weird fallout from a full package bulk build with a jpeg update I'm working on. Rebuilding epeg just ended with the same error. I have now removed all packages from that machine, rebuilt and reinstalled libjpeg from scratch, and now epeg builds. Very strange. -- Christian "naddy" Weisgerber [email protected]
