threads / discuss / 6449

SEGV in git-apply

Subject: SEGV in git-apply

## tl;dr

7 messages between Jan 20, 2007 and Jan 21, 2007.

replies: 6people: 3as markdown or json

Martin Waitz· Jan 20, 2007, 01:48 UTC · lore
hoi :)

I just hit this bug in git-apply, while executing it in a subdirectory. At the moment I am using some mixture of next and some own changes (mostly submodule support, should not affect this bug).

I am afraid that I don't have the time to fix it myself atm, so I am simply attaching the backtrace here:

Program received signal SIGSEGV, Segmentation fault.
0x080500fa in apply_patch (fd=5, filename=0xbf821b4d "/tmp/404241.patch",
    inaccurate_eof=<value optimized out>) at builtin-apply.c:2504
2504                    if (pathlen <= prefix_length ||
(gdb) bt
#0  0x080500fa in apply_patch (fd=5, filename=0xbf821b4d "/tmp/404241.patch",
    inaccurate_eof=<value optimized out>) at builtin-apply.c:2504
#1  0x0805147a in cmd_apply (argc=3, argv=0xbf8201e4, prefix=0x80e75c6 "src/")
    at builtin-apply.c:2719
#2  0x0804ab55 in handle_internal_command (argc=3, argv=0xbf8201e4,
    envp=<value optimized out>) at git.c:304
#3  0x0804b1f8 in main (argc=3, argv=Cannot access memory at address 0x8
) at git.c:341
-- 
Martin Waitz
Johannes Schindelin· Jan 20, 2007, 02:14 UTC · re: Martin Waitz · lore

Re: SEGV in git-apply

Hi,
On Sat, 20 Jan 2007, Martin Waitz wrote:
> Program received signal SIGSEGV, Segmentation fault.
> 0x080500fa in apply_patch (fd=5, filename=0xbf821b4d "/tmp/404241.patch",
>     inaccurate_eof=<value optimized out>) at builtin-apply.c:2504
> 2504                    if (pathlen <= prefix_length ||

It seems that the culprit is not this line (you have an optimized binary, so that is expected).

My guess is that your patch does not contain a new name, neither an old name.

It could also be that the filename has length 0.
But without that patch I cannot tell.

Ciao, Dscho

Martin Waitz· Jan 20, 2007, 18:36 UTC · re: Johannes Schindelin · lore

Re: SEGV in git-apply

hoi :)
On Sat, Jan 20, 2007 at 03:14:46AM +0100, Johannes Schindelin wrote:
> But without that patch I cannot tell.
patch was this one:

------------------ 8< ------------------------------- --- src/callbacks.cc 2006-11-02 10:16:50.000000000 +0100 +++ src/callbacks.cc 2007-01-17 20:47:00.000000000 +0100

@@ -12890,12 +12890,11 @@ void on_unknown_edit_optionmenu_sign_cha
 	}
 }
 
-gboolean on_key_press_event(GtkWidget*, GdkEventKey *event, gpointer) {
+gboolean on_key_press_event(GtkWidget *o, GdkEventKey *event, gpointer) {
 	if(!GTK_WIDGET_HAS_FOCUS(expression) && (event->keyval > GDK_Hyper_R || event->keyval < GDK_Shift_L)) {
-		bool return_val = FALSE;
 		GtkWidget *w = gtk_window_get_focus(GTK_WINDOW(glade_xml_get_widget (main_glade, "main_window")));
-		if(w) g_signal_emit_by_name((gpointer) w, "key_press_event", event, &return_val);
-		if(return_val) return TRUE;
+		if(gtk_bindings_activate_event(GTK_OBJECT(o), event)) return TRUE;
+		if(w && gtk_bindings_activate_event(GTK_OBJECT(w), event)) return TRUE;
 		focus_keeping_selection();
 	}
 	return FALSE;
------------------ 8< -------------------------------

git-apply inside the "src" directory segfaulted.

-- 
Martin Waitz
Johannes Schindelin· Jan 21, 2007, 01:17 UTC · re: Martin Waitz · lore

[PATCH] apply --cached: fix crash in subdirectory

The static variable "prefix" was shadowed by an unused parameter of the same name. In case of execution in a subdirectory, the static variable was accessed, leading to a crash.

Signed-off-by: Knoppix User <knoppix@zweitrechner.(none)>
---
	On Sat, 20 Jan 2007, Martin Waitz wrote:
	
	> git-apply inside the "src" directory segfaulted.
	Well, that was not the complete truth now, was it? Cannily, you 
	avoided mentioning the use of the "--cached" argument...
	But as you see, your evil plan failed ;-)
 builtin-apply.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/builtin-apply.c b/builtin-apply.c
index 54fd2cb..ef927f8 100644
--- a/builtin-apply.c
+++ b/builtin-apply.c
@@ -2589,7 +2589,7 @@ static int git_apply_config(const char *var, const char *value)
 }
 
 
-int cmd_apply(int argc, const char **argv, const char *prefix)
+int cmd_apply(int argc, const char **argv, const char *prefix2)
 {
 	int i;
 	int read_stdin = 1;

-- 
1.5.0.rc1.gd85c
Martin Waitz· Jan 21, 2007, 14:40 UTC · re: Johannes Schindelin · lore

Re: [PATCH] apply --cached: fix crash in subdirectory

hoi :)
On Sun, Jan 21, 2007 at 02:17:19AM +0100, Johannes Schindelin wrote:
Show 6 quoted lines
> 	On Sat, 20 Jan 2007, Martin Waitz wrote:
> 	
> 	> git-apply inside the "src" directory segfaulted.
> 
> 	Well, that was not the complete truth now, was it? Cannily, you 
> 	avoided mentioning the use of the "--cached" argument...

You are right. I forgot to mention that in the bug report. Well, I was in a hurry...

Thanks for being faster in fixing the bug than I am in getting some spare time!

> 	But as you see, your evil plan failed ;-)

Damn! I'll need a better plan next time...

;-)
-- 
Martin Waitz
Johannes Schindelin· Jan 21, 2007, 18:03 UTC · re: Rogan Dawes · lore

Re: [PATCH] apply --cached: fix crash in subdirectory

Hi,
On Sun, 21 Jan 2007, Rogan Dawes wrote:
Show 6 quoted lines
> Johannes Schindelin wrote:
> 
> > 
> > Signed-off-by: Knoppix User <knoppix@zweitrechner.(none)>
> 
> Really?
No :-)
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
You got me there; I use this machine rarely for development.

Ciao, Dscho

← back to recent threads