From: Jakub Jelinek <jakub@redhat.com>
To: Ulrich Drepper <drepper@redhat.com>
Cc: Glibc hackers <libc-hacker@sources.redhat.com>
Subject: [PATCH] Fix LD_DEBUG=statistics
Date: Sat, 13 Nov 2004 18:01:00 -0000 [thread overview]
Message-ID: <20041113175819.GF3536@sunsite.ms.mff.cuni.cz> (raw)
Hi!
LD_DEBUG=statistics someprog leads to segfaults with current glibc.
The problem is that print_statistics doesn't check for NULL _ns_loaded
namespaces and happily dereferences it.
While looking into it, I have noticed the num_relative_relocations
computation doesn't match what do-rel.h is actually doing, particularly
on !ELF_MACHINE_REL_RELATIVE RELA architectures it is processing relative
relocations even if l_addr == 0, but the library is not prelinked.
2004-11-13 Jakub Jelinek <jakub@redhat.com>
* elf/rtld.c (print_statistics): Avoid segfaults if not all namespaces
are used. Fix computation of num_relative_relocations on RELA
architectures other than IA-64 and Alpha.
--- libc/elf/rtld.c.jj 2004-11-09 12:26:41.000000000 +0100
+++ libc/elf/rtld.c 2004-11-13 18:56:32.574028935 +0100
@@ -2336,19 +2336,29 @@ print_statistics (hp_timing_t *rtld_tota
unsigned long int num_relative_relocations = 0;
for (Lmid_t ns = 0; ns < DL_NNS; ++ns)
{
+ if (GL(dl_ns)[ns]._ns_loaded == NULL)
+ continue;
+
struct r_scope_elem *scope = &GL(dl_ns)[ns]._ns_loaded->l_searchlist;
for (unsigned int i = 0; i < scope->r_nlist; i++)
{
struct link_map *l = scope->r_list [i];
- if (!l->l_addr)
- continue;
-
- if (l->l_info[VERSYMIDX (DT_RELCOUNT)])
+ if (l->l_addr != 0 && l->l_info[VERSYMIDX (DT_RELCOUNT)])
num_relative_relocations
+= l->l_info[VERSYMIDX (DT_RELCOUNT)]->d_un.d_val;
- if (l->l_info[VERSYMIDX (DT_RELACOUNT)])
+#ifndef ELF_MACHINE_REL_RELATIVE
+ /* Relative relocations are processed on these architectures if
+ library is loaded to different address than p_vaddr or
+ if not prelinked. */
+ if ((l->l_addr != 0 || !map->l_info[VALIDX(DT_GNU_PRELINKED)])
+ && l->l_info[VERSYMIDX (DT_RELACOUNT)])
+#else
+ /* On e.g. IA-64 or Alpha, relative relocations are processed
+ only if library is loaded to different address than p_vaddr. */
+ if (l->l_addr != 0 && l->l_info[VERSYMIDX (DT_RELACOUNT)])
+#endif
num_relative_relocations
+= l->l_info[VERSYMIDX (DT_RELACOUNT)]->d_un.d_val;
}
Jakub
next reply other threads:[~2004-11-13 18:01 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-11-13 18:01 Jakub Jelinek [this message]
2004-11-13 18:58 ` Ulrich Drepper
-- strict thread matches above, loose matches on Subject: below --
2002-10-24 12:03 Jakub Jelinek
2002-10-25 3:38 ` Roland McGrath
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=20041113175819.GF3536@sunsite.ms.mff.cuni.cz \
--to=jakub@redhat.com \
--cc=drepper@redhat.com \
--cc=libc-hacker@sources.redhat.com \
/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).