| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
This fixes bug #85.
|
|
|
|
| |
This fixes bug #88.
|
|
|
|
| |
This fixes bug #89.
|
|
|
|
|
|
|
| |
We need to cleanup the channels first cause we call ssh_channel_close()
on the channels which still require a working socket and poll context.
Thanks to sh4rm4!
|
| |
|
|
|
|
| |
(cherry picked from commit ceb8072b34a581eb72ed43f7f7ce78fd9b0ea708)
|
|
|
|
| |
(cherry picked from commit 656fd60110b73b3fc56c8c407b12a68be7ca67ff)
|
|
|
|
| |
(cherry picked from commit 6092596199e8ebfbbf6e6ba48b7ec6ca0d85fa13)
|
|
|
|
| |
(cherry picked from commit 22f607649d7dc530df84b7b90781ce0b88153095)
|
|
|
|
| |
(cherry picked from commit b5c4b090da254c1ad3689983b0cb999b09aa3394)
|
|
|
|
| |
(cherry picked from commit 61d032fc03e0055c859931f466bc75fbdf36385a)
|
|
|
|
| |
(cherry picked from commit 280ce3fe937c96899732814928c8ecfa9f9ae431)
|
|
|
|
|
|
|
|
|
|
|
| |
* src/socket.c (ssh_socket_pollcallback): Factor some code out to ...
(ssh_socket_set_connecting): New.
* include/libssh/socket.h (ssh_socket_set_connecting): Add prototype.
* src/client.c (ssh_connect): Use new function for a socket set by
SSH_OPTIONS_FD.
Signed-off-by: Werner Koch <wk@gnupg.org>
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
|
|
|
|
| |
(cherry picked from commit 782b2e37c6a4e0254e2b596b9182043c5ebabd75)
|
|
|
|
|
| |
resolves: #84
(cherry picked from commit a92c97b2e17715c1b3cdd693d14af6c3311d8e44)
|
|
|
|
| |
(cherry picked from commit 8b8d9dc83af81dfaec22bbb6710083e79f00c470)
|
|
|
|
|
|
|
|
| |
It is possible that we get unrelated packets while waiting for
termination, thus waiting indefinitely. As a workaround we have to
check the user-supplied timeout.
Also cleaned up ssh_blocking_flush, which was using the timeout in a
bogus manner (resetting the timeout after each check).
|
|
|
|
| |
(cherry picked from commit 2f861a858be9f2fe1f55a0fbfc47d732e3b3b31b)
|
|
|
|
|
| |
Reply to unknown global messages as required by the RFC. Therefore
keepalive@openssh.com style messages should get treated in a sane way.
|
|
|
|
| |
Introduced with the last commit.
|
|
|
|
|
|
|
|
| |
(cherry picked from commit da954c2c5ee85bcbe2f5ad9f507a6306b0f5f8b9)
Conflicts:
src/keyfiles.c
|
|
|
|
| |
(cherry picked from commit 2cc95e1e088553a94b0a8e07c458b1bc0a60d9b1)
|
|
|
|
| |
(cherry picked from commit c84380bad55dba3cb06486dc7f19c0ea6705a060)
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
(cherry picked from commit 3eece8ac0b107a7df8d95325ef17ed19d6429e75)
Conflicts:
src/channels.c
src/channels1.c
|
|
|
|
|
|
|
|
|
| |
(cherry picked from commit 6d8bb956c5caa48c2aba6713f067224650c3c1e1)
Conflicts:
src/channels.c
src/session.c
|
|
|
|
|
|
|
| |
This code was weird in the first place. I suspect my change will break something else
(probably the appcode that needed it). ssh_poll_ctx_free is not a good
place to send exception callbacks imho.
(cherry picked from commit b5351f2809140921076ef54cc6092b543b5199d2)
|
|
|
|
|
|
| |
Provides me a 3x performance boost for async sftp, 5x for sync sftp
(on localhost)
(cherry picked from commit 17ae216340fea65edbfbb59789121e6e4af73f20)
|
|
|
|
|
|
|
| |
(cherry picked from commit a2c94abb92ead1503a0d6284609af41e53bef402)
Conflicts:
src/channels.c
|
|
|
|
| |
(cherry picked from commit 7363b29427023bb2bbcec0bb1783e869015ff664)
|
|
|
|
| |
(cherry picked from commit 99e6fde751365f0f3508ce9e1899d28ca4c3edb3)
|
|
|
|
| |
(cherry picked from commit 66188f1af84852b9d9fc09acf26454c4cfacefef)
|
|
|
|
|
|
|
|
|
| |
sftp_async_read() and sftp_async_read_begin() assume that the whole read
will be successful but when this is not the case, the offset will be
wrong.
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
(cherry picked from commit d1df255df46b0b59ce5abfa1b2a4b971ebb4fb64)
|
|
|
|
| |
(cherry picked from commit e5e673bafe18032df3b6b7634410a6795e809f2a)
|
|
|
|
|
| |
The server will not return SSH_AUTH_METHOD_NONE.
(cherry picked from commit 9c376dd913b40bbcf9e0fc528f53e02c104b6c58)
|
|
|
|
| |
(cherry picked from commit 13227714f25b57b95594f33d395bc3cea0b7fe8e)
|
|
|
|
|
|
| |
This happened due to the use of the buggy and obsolete timeout
funtions.
(cherry picked from commit 7949f2cdc6eb1ac0f076e06e6a3fa47540a9b870)
|
|
|
|
| |
(cherry picked from commit 2f878736427f5fd846f8ebb0c834e6a65bf8a638)
|
|
|
|
|
|
| |
Signed-off-by: Mark Riordan <mriordan@ipswitch.com>
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
(cherry picked from commit 6c45d6dc0196a2846313b588c0e05996813ab3f8)
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There where two issues with ssh_scp_write:
1) It did not write a status message after the last write and OpenSSH
would then give up after the write finished.
2) OpenSSH would sometimes write a status message, after near ends write.
If scp_write didn't handle it, and subsequent status message. The remote
window would shrink to zero and ssh_channel_write would start returning 0.
Signed-off-by: Einar Floystad Dorum <einarfd@mailthief.com>
Signed-off-by: Andreas Schneider <asn@cryptomilk.org>
(cherry picked from commit 01c4b713dc57d16612c3b61e1857fb982623dd12)
|
|
|
|
| |
(cherry picked from commit 35686b4822c8a18d1d6715221be91ee67e563bad)
|
|
|
|
| |
(cherry picked from commit 7ccd9c31b33798824b27e4af095b52b9026a1f98)
|
|
|
|
|
|
| |
-2 now means to use the timeout specified in options. It wasn't used
earlier and poll only knows -1 and 0 anyway for special meanings.
(cherry picked from commit af85337f5ff052437fa5cf117f3d065488640660)
|
|
|
|
|
|
|
|
|
| |
We need this that we don't end up in and infinite poll loop.
(cherry picked from commit 563fbe4de8ee090b40b50415a86f9a3da16f46b0)
Conflicts:
src/poll.c
|
|
|
|
|
|
| |
It only worked if the first channel in the list was equivalent to we
were looking for.
(cherry picked from commit 39f962c91eb4575a65edc7d984ce3f1a699097b8)
|