This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
On Tue, Jan 9, 2018 at 11:33 AM, Florian Weimer <fweimer@redhat.com> wrote: > On 01/09/2018 05:29 PM, Zack Weinberg wrote: >> >> On Mon, Jan 8, 2018 at 11:59 AM, Zack Weinberg <zackw@panix.com> wrote: >>> >>> This patch removes the #include of libio.h from the installed stdio.h, >> >> ... >>> >>> This passes the testsuite on x86-64-linux-gnu. A build-many-glibcs >>> cycle is in progress. >> >> >> No new failures from build-many-glibcs. > > What about material object file changes? There are some changes to installed stripped libraries, but I believe all of them to be immaterial. The removal of the #defines for getc and putc means that several of the ancillary libraries and programs change from calling _IO_getc@GLIBC_2.2.5 or _IO_putc@GLIBC_2.2.5 to calling getc@GLIBC_2.2.5 or putc@GLIBC_2.2.5, and this causes the PLT to get reorganized and many section base addresses to shift a little bit. diffoscope doesn't handle this case very well, and shows an enormous diff for e.g. the text segment of libresolv, but I *think* the code itself didn't change, it just moved a little in memory. Also, removing lines from the top of vfprintf.c caused all the assertions in that file to have different line numbers. And I don't know how to freeze the build-id notes. That appears to be everything. For interest, the complete output of `diffoscope --exclude-directory-metadata --exclude '*.h'` is attached.
Attachment:
glibc-diff.txt.gz
Description: GNU Zip compressed data
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |