From: "H.J. Lu" <hjl.tools@gmail.com>
To: binutils@sourceware.org, Alan Modra <amodra@gmail.com>
Subject: Re: [GOLD] testsuite tweaks for powerpc
Date: Thu, 06 Sep 2012 04:26:00 -0000 [thread overview]
Message-ID: <CAMe9rOr6aP0cABE1+jbHLmYYqZ-O_s64nBMYhqVrpmQr6Lz7rQ@mail.gmail.com> (raw)
In-Reply-To: <20120905034538.GS3159@bubble.grove.modra.org>
On Tue, Sep 4, 2012 at 8:45 PM, Alan Modra <amodra@gmail.com> wrote:
> Various test tweaks for powerpc.
> - Non-pic shared libs don't work.
> - Adding synthetic syms lets the test see the ordering of code sections
> which is what the tests are checking, rather than the ordering of
> .opd entries.
> - memory_test assumes 4k pages
> - justsyms_exec looks at the address of a function, which on powerpc64
> is the descriptor in .opd, a data section.
> - .toc input sections on powerpc64 need to be part of the .got output
> section.
>
> Tested on x86_64 too. OK?
>
> * configure.ac (FN_PTRS_IN_SO_WITHOUT_PIC): False for powerpc.
> * configure: Regenerate.
> * testsuite/Makefile.am (final_layout.stdout): Pass --synthetic to nm.
> (plugin_final_layout.stdout): Likewise.
> (memory_test): Set page sizes to 0x1000.
> * testsuite/Makefile.in: Regenerate.
> * testsuite/discard_locals_test.sh: Add FIXME comment.
> * testsuite/justsyms_exec.c: Disable function test for powerpc64.
> * testsuite/pr14265.t: Add .got output section statement.
> * testsuite/script_test_2.t: Likewise.
> * testsuite/script_test_3.t: Likewise.
> * testsuite/script_test_4.t: Likewise.
> * testsuite/script_test_5.t: Likewise.
> * testsuite/script_test_6.t: Likewise.
> * testsuite/script_test_7.t: Likewise.
> * testsuite/script_test_9.t: Likewise.
>
> Index: gold/testsuite/script_test_3.t
> ===================================================================
> RCS file: /cvs/src/src/gold/testsuite/script_test_3.t,v
> retrieving revision 1.5
> diff -u -p -r1.5 script_test_3.t
> --- gold/testsuite/script_test_3.t 15 Sep 2010 17:39:57 -0000 1.5
> +++ gold/testsuite/script_test_3.t 5 Sep 2012 03:20:05 -0000
> @@ -35,6 +35,7 @@ SECTIONS
> . = ALIGN(0x100);
> .dynamic : { *(.dynamic) } :data :dynamic
> .data : { *(.data) } :data
> + .got : { *(.got .toc) }
> .tdata : { *(.tdata*) } :data :tls
> .tbss : { *(.tbss*) } :data :tls
> . += 0x100000;
> Index: gold/testsuite/script_test_4.t
> ===================================================================
> RCS file: /cvs/src/src/gold/testsuite/script_test_4.t,v
> retrieving revision 1.2
> diff -u -p -r1.2 script_test_4.t
> --- gold/testsuite/script_test_4.t 15 Sep 2010 17:39:57 -0000 1.2
> +++ gold/testsuite/script_test_4.t 5 Sep 2012 03:20:05 -0000
> @@ -37,6 +37,7 @@ SECTIONS
> . = ALIGN(0x100);
> .dynamic : { *(.dynamic) }
> .data : { *(.data) }
> + .got : { *(.got .toc) }
> . += 0x100000;
> . = ALIGN(0x100);
> .bss : { *(.bss) }
This change generates strange section layout on x86-64:
[28] .got.plt PROGBITS 0000000010101d40 001d40
000060 00 WA 0 0 8
[29] .tdata PROGBITS 0000000010101da0 001da0
00002c 00 WAT 0 0 8
[30] .tbss NOBITS 0000000010101dd0 001dd0
00002c 00 WAT 0 0 8
[31] .got PROGBITS 0000000010101dd0 001dd0
000008 00 WA 0 0 8
The executable runs. But on x32, I got
[28] .got.plt PROGBITS 10101b20 001b20 000060 00 WA 0 0 8
[29] .tdata PROGBITS 10101b80 001b80 000024 00 WAT 0 0 8
[30] .tbss NOBITS 10101ba4 001ba4 000020 00 WAT 0 0 4
[31] .got PROGBITS 10101ba8 001ba8 000008 00 WA 0 0 8
and
[hjl@gnu-6 testsuite]$ ./tls_script_test
Segmentation fault
[hjl@gnu-6 testsuite]$
Total failures on x32 are:
/bin/sh: line 1: 30456 Segmentation fault "$tst" >
script_test_3.log-t 2>&1FAIL: script_test_3
PASS: binary_test
PASS: thin_archive_test_1
/bin/sh: line 1: 30493 Segmentation fault "$tst" >
tls_phdrs_script_test.log-t 2>&1
FAIL: tls_phdrs_script_test
/bin/sh: line 1: 30486 Segmentation fault "$tst" >
tls_script_test.log-t 2>&1
PASS: relro_now_test
FAIL: tls_script_test
I think they work by pure luck on other targets.
This patch fixes those 3 failures on x32. I didn't
change other linker scripts since they don't cause
any failures.
--
H.J.
---
diff --git a/gold/testsuite/script_test_3.t b/gold/testsuite/script_test_3.t
index 9c75194..accd055 100644
--- a/gold/testsuite/script_test_3.t
+++ b/gold/testsuite/script_test_3.t
@@ -36,6 +36,7 @@ SECTIONS
.dynamic : { *(.dynamic) } :data :dynamic
.data : { *(.data) } :data
.got : { *(.got .toc) }
+ .got.plt : { *(.got.plt) }
.tdata : { *(.tdata*) } :data :tls
.tbss : { *(.tbss*) } :data :tls
. += 0x100000;
diff --git a/gold/testsuite/script_test_4.t b/gold/testsuite/script_test_4.t
index f5569c5..3ba5e93 100644
--- a/gold/testsuite/script_test_4.t
+++ b/gold/testsuite/script_test_4.t
@@ -38,6 +38,7 @@ SECTIONS
.dynamic : { *(.dynamic) }
.data : { *(.data) }
.got : { *(.got .toc) }
+ .got.plt : { *(.got.plt) }
. += 0x100000;
. = ALIGN(0x100);
.bss : { *(.bss) }
next prev parent reply other threads:[~2012-09-06 4:26 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-09-05 3:46 Alan Modra
2012-09-05 4:33 ` Ian Lance Taylor
2012-09-06 4:26 ` H.J. Lu [this message]
2012-09-06 4:40 ` Ian Lance Taylor
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAMe9rOr6aP0cABE1+jbHLmYYqZ-O_s64nBMYhqVrpmQr6Lz7rQ@mail.gmail.com \
--to=hjl.tools@gmail.com \
--cc=amodra@gmail.com \
--cc=binutils@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).