From: Johannes Schindelin <hidden> Date: 2016-06-15 22:47:15
Hi,
On Mon, 17 Aug 2009, Ali Polatel wrote:
Here's what gdb has to say about it:
[...]
Here's what valgrind has to say about it (with a current 'next' +
patches):
==25434== Invalid read of size 8
==25434== at 0x407433: process_object_response (http-walker.c:91)
==25434== by 0x405713: finish_active_slot (http.c:657)
==25434== by 0x40448F: process_curl_messages (http.c:119)
==25434== by 0x40546A: step_active_slots (http.c:571)
==25434== by 0x4080E8: fetch_object (http-walker.c:476)
==25434== by 0x408316: fetch (http-walker.c:526)
==25434== by 0x42876C: loop (walker.c:176)
==25434== by 0x428C65: walker_fetch (walker.c:287)
==25434== by 0x40401F: main (remote-curl.c:111)
Ciao,
Dscho
From: Johannes Schindelin <hidden> Date: 2016-06-15 22:47:15
Hi,
On Mon, 17 Aug 2009, Johannes Schindelin wrote:
On Mon, 17 Aug 2009, Ali Polatel wrote:
quoted
Here's what gdb has to say about it:
[...]
Here's what valgrind has to say about it (with a current 'next' +
patches):
==25434== Invalid read of size 8
==25434== at 0x407433: process_object_response (http-walker.c:91)
==25434== by 0x405713: finish_active_slot (http.c:657)
==25434== by 0x40448F: process_curl_messages (http.c:119)
==25434== by 0x40546A: step_active_slots (http.c:571)
==25434== by 0x4080E8: fetch_object (http-walker.c:476)
==25434== by 0x408316: fetch (http-walker.c:526)
==25434== by 0x42876C: loop (walker.c:176)
==25434== by 0x428C65: walker_fetch (walker.c:287)
==25434== by 0x40401F: main (remote-curl.c:111)
Oops, forgot the more important part:
==25434== Address 0x87a7bd0 is 0 bytes inside a block of size 64 free'd
==25434== at 0x4C265AF: free (vg_replace_malloc.c:323)
==25434== by 0x4075F5: release_object_request (http-walker.c:128)
==25434== by 0x408022: abort_object_request (http-walker.c:452)
==25434== by 0x4080D7: fetch_object (http-walker.c:470)
==25434== by 0x408316: fetch (http-walker.c:526)
==25434== by 0x42876C: loop (walker.c:176)
==25434== by 0x428C65: walker_fetch (walker.c:287)
==25434== by 0x40401F: main (remote-curl.c:111)
Seems that an object request is aborted, but the slot, and therefore the
callback, is called nevertheless. Tay, does that ring a bell?
Ciao,
Dscho
From: Lars Hjemli <hidden> Date: 2016-06-15 22:47:15
On Mon, Aug 17, 2009 at 15:56, Ali Polatel[off-list ref] wrote:
$subject.
git version 1.6.4
Here's what gdb has to say about it:
2456 alip@harikalardiyari> gdb --args git clone http://git.savannah.gnu.org/cgit/xboard.git
...
Getting pack 06483273097cbac210f10a4bd43324ae660053e6
which contains 74e24bdc2ec3f275da63ca1396a773e7043cb9e9
Program received signal SIGSEGV, Segmentation fault.
From: Tay Ray Chuan <hidden> Date: 2016-06-15 22:47:19
Hi,
On Mon, Aug 17, 2009 at 10:22 PM, Johannes Schindelin[off-list ref] wrote:
Seems that an object request is aborted, but the slot, and therefore the
callback, is called nevertheless. Tay, does that ring a bell?
thanks Johannes, your diagnosis was a vital clue.
Ali, could you see if this patch fixes it for you? On my side, I had
some difficulty reproducing your problem reliably (it happened
sometimes but not on other times).
--
Cheers,
Ray Chuan
-- >8 --
Subject: [PATCH] http.c: set slot callback members to NULL when releasing object
Set the members callback_func and callback_data of freq->slot to NULL
when releasing a http_object_request. release_active_slot() is also
invoked on the slot to remove the curl handle associated with the slot
from the multi stack (CURLM *curlm in http.c).
These prevent the callback function and data from being used in http
methods (like http.c::finish_active_slot()) after a
http_object_request has been free'd.
Signed-off-by: Tay Ray Chuan <redacted>
---
http.c | 7 ++++++-
1 files changed, 6 insertions(+), 1 deletions(-)
From: Ali Polatel <hidden> Date: 2016-06-15 22:47:19
Tay Ray Chuan yazmış:
Hi,
On Mon, Aug 17, 2009 at 10:22 PM, Johannes Schindelin[off-list ref] wrote:
quoted
Seems that an object request is aborted, but the slot, and therefore the
callback, is called nevertheless. Tay, does that ring a bell?
thanks Johannes, your diagnosis was a vital clue.
Ali, could you see if this patch fixes it for you? On my side, I had
some difficulty reproducing your problem reliably (it happened
sometimes but not on other times).
It works, I don't get any segfaults after applying this patch.
--
Regards,
Ali Polatel
From: Johannes Schindelin <hidden> Date: 2016-06-15 22:47:19
Hi,
On Wed, 26 Aug 2009, Ali Polatel wrote:
Tay Ray Chuan yazmış:
quoted
On Mon, Aug 17, 2009 at 10:22 PM, Johannes Schindelin[off-list ref] wrote:
quoted
Seems that an object request is aborted, but the slot, and therefore
the callback, is called nevertheless. Tay, does that ring a bell?
thanks Johannes, your diagnosis was a vital clue.
Ali, could you see if this patch fixes it for you? On my side, I had
some difficulty reproducing your problem reliably (it happened
sometimes but not on other times).
It works, I don't get any segfaults after applying this patch.
Great!
But why did you drop me from the Cc: list? It's not every day that I can
pay that close attention to the mails I get; mails which are not addressed
to me directly fall off the plate on other days...
Ciao,
Dscho
From: Tay Ray Chuan <hidden> Date: 2016-06-15 22:47:20
Hi,
2009/8/26 Ali Polatel [off-list ref]:
It works, I don't get any segfaults after applying this patch.
Junio, I hope you don't mind me asking but why hasn't this patch been
accepted? It addresses a pretty severe problem, and the sooner users
have it the better.
--
Cheers,
Ray Chuan