#351 closed defect (fixed)
msn Error: Challenge response invalid
Reported by: | Owned by: | wilmer | |
---|---|---|---|
Priority: | major | Milestone: | 1.2 |
Component: | MSN | Version: | devel |
Keywords: | challenge response | Cc: | |
IRC client+version: | irssi 0.8.13-svn (20071223 1648) | Operating System: | MacOS |
OS version/distro: | MacOS 10.4.11 |
Description
After "account on msn" I get the following message:
13:24:37 <@root> msn - Logging in: Connecting 13:24:37 <@root> msn - Logging in: Connected to server, waiting for reply 13:24:38 <@root> msn - Logging in: Transferring to other server 13:24:38 <@root> msn - Logging in: Connected to server, waiting for reply 13:24:40 <@root> msn - Logging in: Authenticated, getting buddy list 13:24:41 <@root> msn - Logging in: Logged in 13:25:05 <@root> msn - Error: Error reported by MSN server: Challenge response invalid 13:25:05 <@root> msn - Signing off..
This bitlbee version bzr-305
Attachments (0)
Change History (10)
comment:1 Changed at 2008-01-17T21:50:58Z by
Priority: | critical → major |
---|
comment:2 Changed at 2008-01-18T07:24:18Z by
Indeed, all the time; also with bzr-307.
Yes, running on OS X, sorry, was confused by web interface.
The tests (you meant running "make check"?) fail here:
[...] make -C tests
- Compiling check.c
check.c:4:19: error: check.h: No such file or directory
comment:3 follow-ups: 7 8 Changed at 2008-01-18T10:56:30Z by
Milestone: | → 1.2 |
---|---|
Operating System: | Public server → MacOS |
Version: | 1.0.3 → devel |
Yeah, that's what I was afraid of. But if you use Fink/Darwinports you might be able to install the "check" port/package to get it to work. Do you maybe have a Jabber/GMail account you can try out? Usually that'll use MD5 too, so if that also fails we can be sure the problem is there.
Another thing you can try: bzr revision 266 still had the original (GPL-incompatible) MD5 code. If that one works and revision 267 doesn't: Gotcha.
If you don't have time for this, I might be able to do some testing on my work MBP as well. I have to get bzr to work on it though. :-/
comment:4 follow-up: 5 Changed at 2008-01-18T10:57:54Z by
Oh yes, one more important detail: Are you using an Intel- or PPC-based Mac?
comment:5 Changed at 2008-01-18T11:28:12Z by
Replying to wilmer:
Oh yes, one more important detail: Are you using an Intel- or PPC-based Mac?
ppc
comment:6 Changed at 2008-01-18T11:50:56Z by
Okay, it must be an endianness issue then. Do you have time to install the check port/pacakge so you can run the unit tests?
comment:7 Changed at 2008-01-18T12:06:34Z by
Replying to wilmer:
Yeah, that's what I was afraid of. But if you use Fink/Darwinports you might be able to install the "check" port/package to get it to work. Do you maybe have a Jabber/GMail account you can try out? Usually that'll use MD5 too, so if that also fails we can be sure the problem is there.
make -C tests
- Linking check
./check Warning: Unable to read configuration file `(null)'. Running suite(s): Util
Nick MD5 ArcFour IRC
socketpair: Protocol not supported socketpair: Protocol not supported
Help User
socketpair: Protocol not supported socketpair: Protocol not supported socketpair: Protocol not supported socketpair: Protocol not supported socketpair: Protocol not supported socketpair: Protocol not supported socketpair: Protocol not supported
Crypting Set
77%: Checks: 44, Failures: 5, Errors: 5 check_md5.c:45:F:Core:check_sums:0: failed check_irc.c:14:F:Core:test_connect:0: Assertion 'g_io_channel_pair(&ch1, &ch2)' failed check_irc.c:32:F:Core:test_login:0: Assertion 'g_io_channel_pair(&ch1, &ch2)' failed check_user.c:10:E:Core:test_user_add:0: (after this point) Received signal 10 (Bus error) check_user.c:19:E:Core:test_user_add_exists:0: (after this point) Received signal 10 (Bus error) check_user.c:48:E:Core:test_user_del_nonexistant:0: (after this point) Received signal 10 (Bus error) check_user.c:40:E:Core:test_user_del:0: (after this point) Received signal 10 (Bus error) check_user.c:53:E:Core:test_user_rename:0: (after this point) Received signal 10 (Bus error) check_crypting.c:11:F:Core:test_check_pass_valid:0: Assertion 'checkpass ("foo", "acbd18db4cc2f85cedef654fccc4a4d8") == 0' failed check_crypting.c:17:F:Core:test_hashpass:0: Assertion 'strcmp(hashpass("foo"), "acbd18db4cc2f85cedef654fccc4a4d8") == 0' failed make[1]: * [all] Error 1 make: * [check] Error 2
Another thing you can try: bzr revision 266 still had the original (GPL-incompatible) MD5 code. If that one works and revision 267 doesn't: Gotcha.
Then pass isn't valid anymore.
But check_crypting passes with 266.
comment:8 Changed at 2008-01-18T12:38:44Z by
Replying to wilmer:
Another thing you can try: bzr revision 266 still had the original (GPL-incompatible) MD5 code. If that one works and revision 267 doesn't: Gotcha.
Ok, set up msn account for 266, then I get:
13:35:32 <@root> msn - Logging in: Connecting 13:35:33 <@root> msn - Logging in: Connected to server, waiting for reply 13:35:33 <@root> msn - Logging in: Transferring to other server 13:35:33 <@root> msn - Logging in: Connected to server, waiting for reply 13:35:36 <@root> msn - Couldn't log in: Error during Passport authentication (HTTP error: 401 Unauthorized) 13:35:36 <@root> msn - Logging in: Signing off..
comment:9 Changed at 2008-01-18T23:53:45Z by
I got a fix here, it was indeed an endianness problem. Will commit it tomorrow.
comment:10 Changed at 2008-01-19T12:42:47Z by
Resolution: | → fixed |
---|---|
Status: | new → closed |
Fixed in changeset:devel,309. Thanks for the report! Surprisingly, nobody else noticed this problem and the broken code was there for two months already.
Do you get that all the time? AFAIK bzr-305 isn't actually running on any public server yet, so I assume you're actually running on OS X. This looks to me like maybe the MD5 code is broken (portability issue). Can you try to run the unit tests to see if they work properly on your machine?