2009/3/27 Daniel Stenberg wrote:
> Yes, that looks like the right thing to do. Will you commit?
Committed !
--
-=[Yang]=-
--
___
libssh2-devel mailing list
libssh2-devel@list
On Fri, 27 Mar 2009, Daniel Johnson wrote:
> Well, I'm not getting segfaults, but I'm still getting test failures:
>
> test 600...[SFTP retrieval]
>
> ** MEMORY FAILURE
> Leak detected: memory still allocated: 188 bytes
> At 30c264, there's 188 bytes.
> allocated by ../../curl/lib/ssh.c:281
>
> wh
Feature Requests item #2719127, was opened at 2009-03-28 10:37
Message generated for change (Tracker Item Submitted) made by benkibbey
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=703945&aid=2719127&group_id=125852
Please note that this message will contain a ful
Hi,
After Daniel remark "Why alloc for 7 bytes? Why alloc for 7 *fixed*
bytes?";
I reread userauth.c code for functions userauth_publickey_fromfile and
userauth_hostbased_fromfile and there is something that IMO is a bit
curious:
In userauth_hostbased_fromfile, we have (comments omitted):
On Sat, 28 Mar 2009, Jean-Louis CHARTON wrote:
> => Here, notice that on failure, session->userauth_host_method is freed.
> Also notice that in userauth_hostbased_fromfile, there is no call to
> free pubkeydata.
> So I believe there is a memory leak in userauth_hostbased_fromfile.
>
> Am I right
On Mar 28, 2009, at 9:54 AM, Daniel Stenberg wrote:
> On Fri, 27 Mar 2009, Daniel Johnson wrote:
>
>> Well, I'm not getting segfaults, but I'm still getting test failures:
>>
>> test 600...[SFTP retrieval]
>>
>> ** MEMORY FAILURE
>> Leak detected: memory still allocated: 188 bytes
>> At 30c264,
Hi Daniel,
I'm glad this help.
Now, maybe you better understand why my code allocated a 7 bytes buffer to
store the method: just because the method is freed later in
userauth_hostbased_fromfile / userauth_publickey_fromfile ...
Maybe a better solution could be to always assume/impose that met
Is libssh2 designed to be used safely in a multi threaded program?
Parts of the code I've read thus far make me think the answer could is
yes.
However, I think I've found at least one non re-entrant peace of code in
openssl.c in _libssh2_rsa_new_private and _libssh2_dsa_new_private. I
mean
the c
Daniel Johnson wrote:
> >> ** MEMORY FAILURE
> >> Leak detected: memory still allocated: 188 bytes
> >> At 30c264, there's 188 bytes.
> >> allocated by ../../curl/lib/ssh.c:281
> >>
> >> which is in libcurl's LIBSSH2_ALLOC_FUNC.
..
> Too bad there's no OS X port of valgrind yet.
You can find out
On Mar 28, 2009, at 9:45 PM, Peter Stuge wrote:
> Daniel Johnson wrote:
** MEMORY FAILURE
Leak detected: memory still allocated: 188 bytes
At 30c264, there's 188 bytes.
allocated by ../../curl/lib/ssh.c:281
which is in libcurl's LIBSSH2_ALLOC_FUNC.
>
> ..
>> Too ba
10 matches
Mail list logo