On 2/6/06, Jeremy Huntwork <[EMAIL PROTECTED]> wrote: > Hello All, > > In talking with Ryan Oliver, there seems to be one final thing that we > can do to our current build which will help stabilize it completely: add > 'make bootstrap' to the gcc build of chapter 6. > <snip> > > Comments?
As I said in another post on this thread, I don't understand the details of using make bootstrap in the context of Ch. 6. So, what I say may be stupid or useless information. I decided to build Ch. 6 gcc three ways: 1. Current LFS (startfile_prefix_spec w/out make bootstrap) 2. Proposed LFS (startfile_prefix_spec w/ make bootstrap) 3. Current DIY (-B/usr/lib/, -L/usr/lib w/out make bootstrap) I installed the three into DESTDIR's and did the same binary diffing routine that I've done for ICA verification. Results: After stripping, they were all the same except for the stdc++ differences that have been in every set of binary diffs I've done. I can provide the results if anyone wants. Now, one thing I will note is that since the glibc in Ch. 5 and Ch. 6 are pretty much the same, linking against the wrong libc is probably not going result in differences. However, I'm going to run the tests again, but building the Ch. 6 glibc optimized since that should be different than the one in /tools. I'm going to add -Wl,-t which prints the library and startfile trace so I can see what's being linked to in the logs. Also, it's hard to tell if the headers from /usr/include are being used in the build. I noticed from cc1 --help that -H prints the name of the headers as they're used. If anyone else has an idea similar to the one above that would highlight when the wrong headers are being used, I'd like to hear it. I was thinking about renaming a header in /tools/include that I know is needed, but I don't know if that would unconditionally break the build. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
