[Bug ld/31970] New: [x86_64] ld testsuite: mark-plt-* tests fail when targeting musl

[awilfox at adelielinux dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=31970

            Bug ID: 31970
           Summary: [x86_64] ld testsuite: mark-plt-* tests fail when
                    targeting musl
           Product: binutils
           Version: 2.42
            Status: UNCONFIRMED
          Keywords: testsuite
          Severity: normal
          Priority: P2
         Component: ld
          Assignee: unassigned at sourceware dot org
          Reporter: awilfox at adelielinux dot org
  Target Milestone: ---
            Target: x86_64-*-linux-musl

Created attachment 15613
  --> https://sourceware.org/bugzilla/attachment.cgi?id=15613&action=edit
Full ld.log from test run

Using:

GNU ld (GNU Binutils) 2.42
gcc (Adelie 13.3.0) 13.3.0

on:

Linux lab-x86_64-lin-2.tpcl.atwilcox.tech 5.15.132-mc6-easy #1 SMP Sun Nov 12
04:16:02 UTC 2023 x86_64 GNU/Linux
Architecture:            x86_64
CPU(s):                  8
Vendor ID:               GenuineIntel
  Model name:            Intel(R) Core(TM) i7-6700T CPU @ 2.80GHz



I am seeing all mark-plt tests fail on x86_64/musl:

Running
/home/awilcox/Code/packages/system/binutils/src/binutils-2.42/ld/testsuite/ld-x86-64/x86-64.exp
...
FAIL: ld-x86-64/mark-plt-1a
FAIL: ld-x86-64/mark-plt-1b
FAIL: ld-x86-64/mark-plt-1c
FAIL: ld-x86-64/mark-plt-1d
FAIL: ld-x86-64/mark-plt-1a-x32
FAIL: ld-x86-64/mark-plt-1b-x32
FAIL: ld-x86-64/mark-plt-1c-x32
FAIL: ld-x86-64/mark-plt-1d-x32
FAIL: Build mark-plt-1.so



When I try to reproduce the mark-plt-1a test from the test suite, I see:

awilcox on lab-x86_64-lin-2
~/Code/packages/system/binutils/src/binutils-2.42/ld % ./ld-new -z norelro 
-L/home/awilcox/Code/packages/system/binutils/src/binutils-2.42/ld/testsuite/ld-x86-64
 -melf_x86_64 -shared -z mark-plt -o tmpdir/dump tmpdir/mark-plt-1.o
awilcox on lab-x86_64-lin-2
~/Code/packages/system/binutils/src/binutils-2.42/ld % ../binutils/readelf 
-drW tmpdir/dump

Dynamic section at offset 0x288 contains 13 entries:
  Tag        Type                         Name/Value
 0x000000006ffffef5 (GNU_HASH)           0x190
 0x0000000000000005 (STRTAB)             0x200
 0x0000000000000006 (SYMTAB)             0x1b8
 0x000000000000000a (STRSZ)              9 (bytes)
 0x000000000000000b (SYMENT)             24 (bytes)
 0x0000000070000000 (DT_X86_64_PLT)      0x230
 0x0000000070000001 (DT_X86_64_PLTSZ)    0x20
 0x0000000070000003 (DT_X86_64_PLTENT)   0x10
 0x0000000000000003 (PLTGOT)             0x13a8
 0x0000000000000002 (PLTRELSZ)           24 (bytes)
 0x0000000000000014 (PLTREL)             RELA
 0x0000000000000017 (JMPREL)             0x210
 0x0000000000000000 (NULL)               0x0

Relocation section '.rela.plt' at offset 0x210 contains 1 entry:
    Offset             Info             Type               Symbol's Value 
Symbol's Name + Addend
00000000000013c0  0000000100000007 R_X86_64_JUMP_SLOT     0000000000000000 bar
+ 240



While we were expecting:

% cat testsuite/ld-x86-64/mark-plt-1a.d
#source: mark-plt-1.s
#as: --64
#ld: -melf_x86_64 -shared -z mark-plt
#readelf: -drW

#...
 0x0000000070000000 \(DT_X86_64_PLT\)      0x1000
 0x0000000070000001 \(DT_X86_64_PLTSZ\)    0x20
 0x0000000070000003 \(DT_X86_64_PLTENT\)   0x10
#...
[0-9a-f ]+R_X86_64_JUMP_SLOT +0+ +bar \+ 1010
#pass



My theory is that this is because there is no glibc dep and therefore no
.gnu_version_r section:

% cat testsuite/ld-x86-64/mark-plt-1a.rd
#...
Version needs section '.gnu.version_r' contains 1 entry:
 Addr: 0x[0-9a-f]+ +Offset: 0x[0-9a-f]+ +Link: +[0-9]+ +\(.dynstr\)
 +0+: Version: 1 +File: libc\.so\.6(|\.1) +Cnt: +[0-9]+
#...
  0x[a-f0-9]+:   Name: GLIBC_2.36  Flags: none  Version: [0-9]+
#pass

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