[ltrace-commits] 01/02: Fix a problem in tracing across fork on PPC64

[Petr Machata]

This is an automated email from the git hooks/post-receive script.

pmachata-guest pushed a commit to branch master
in repository ltrace.

commit 35742523e3daa0e59de0c1c3fdd8e5ff52891967
Author: Petr Machata <pmachata at redhat.com>
Date:   Thu Jan 9 23:41:50 2014 +0100

    Fix a problem in tracing across fork on PPC64
    
    In order to avoid single-stepping through large portions of the
    dynamic linker, ltrace remembers at which address the instruction that
    resolved a PLT slot is.  It then puts a breakpoint to this address so
    that it can fast-forward to that address next time it needs to catch a
    PLT slot being resolved.
    
    When a process is cloned, the pointer to this breakpoint is simply
    copied over to the new process, instead of being looked up in the new
    process structures.  This patches fixes this.
---
 sysdeps/linux-gnu/ppc/plt.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/sysdeps/linux-gnu/ppc/plt.c b/sysdeps/linux-gnu/ppc/plt.c
index 3ec1397..8715da6 100644
--- a/sysdeps/linux-gnu/ppc/plt.c
+++ b/sysdeps/linux-gnu/ppc/plt.c
@@ -1,6 +1,6 @@
 /*
  * This file is part of ltrace.
- * Copyright (C) 2012,2013 Petr Machata, Red Hat Inc.
+ * Copyright (C) 2012,2013,2014 Petr Machata, Red Hat Inc.
  * Copyright (C) 2004,2008,2009 Juan Cespedes
  * Copyright (C) 2006 Paul Gilliam
  *
@@ -1157,6 +1157,18 @@ int
 arch_process_clone(struct process *retp, struct process *proc)
 {
 	retp->arch = proc->arch;
+
+	if (retp->arch.dl_plt_update_bp != NULL) {
+		/* Point it to the corresponding breakpoint in RETP.
+		 * It must be there, this part of PROC has already
+		 * been cloned to RETP.  */
+		retp->arch.dl_plt_update_bp
+			= address2bpstruct(retp,
+					   retp->arch.dl_plt_update_bp->addr);
+
+		assert(retp->arch.dl_plt_update_bp != NULL);
+	}
+
 	return 0;
 }
 

-- 
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/collab-maint/ltrace.git