[Bug ld/29235] Some glibc tests crash in DT_RELR relocation on powerpc64le

[amodra at gmail dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=29235

Alan Modra <amodra at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at sourceware dot org   |amodra at gmail dot com
             Status|NEW                         |ASSIGNED

--- Comment #7 from Alan Modra <amodra at gmail dot com> ---
Mainline binutils results in what looks like a good binary:
  [18] .init_array       INIT_ARRAY      0000000000245db8 245db8 000010 08  WA 
0   0  8
Relocation section '.relr.dyn' at offset 0x16d8 contains 863 entries:
  40670 offsets
0000000000245db8

Compare with dt-relr disabled:
  [17] .init_array       INIT_ARRAY      0000000000335de8 335de8 000010 08  WA 
0   0  8
Relocation section '.rela.dyn' at offset 0x10a0 contains 40683 entries:
    Offset             Info             Type               Symbol's Value 
Symbol's Name + Addend
0000000000335de8  0000000000000016 R_PPC64_RELATIVE                         
f0510

In both cases first relative reloc is on first entry of .init_array.

Tulio likely identified the problem in comment 6.  I can't download
binutils-2.38-14.fc37.src.rpm at the moment to check (Sorry! This service is
currently unavailable), but the log on
https://koji.fedoraproject.org/koji/buildinfo?buildID=1971779 would indicate
the patch is missing.  I see a rebase on Feb 9 and no mention of it later.  The
relr fix landed on the sourceware binutils-2.38 branch on Mar 13.

-- 
You are receiving this mail because:
You are on the CC list for the bug.