4cd723fb7a
Fix crash on timed out mount-nfs calls.
...
If wait_for_nfs_reply() times out, nfs_mount can return with RPCs
still pending. In that case when the RPCs complete (perhaps because
someone calls destroy_context()), the callbacks run, and private_data
is pointing at what was the stack-allocated cb_data structure. Stack
smashing and segfaulty fun ensue.
Fix by ensuring no RPCs are pending before returning from nfs_mount()
by disconnecting on errors.
2017-05-11 15:19:01 -06:00
14adfbfc4b
Merge pull request #211 from earlchew/issues-210
...
Check result of FSINFO before using result in nfs_mount_10_cb()
2017-05-09 22:26:23 -07:00
d8d9e5556b
Merge pull request #214 from earlchew/issues-206-01
...
Return an error if either readmax or writemax are less than NFSMAXDATA2
2017-05-09 22:07:18 -07:00
2f041424d5
Return an error if either readmax or writemax are less than NFSMAXDATA2
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-09 21:55:03 -07:00
63d6679132
Check result of FSINFO before using result in nfs_mount_10_cb()
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-09 21:35:30 -07:00
bf7d983812
Merge pull request #208 from earlchew/issues-200
...
Latch error return from nfs_mount_8_cb()
2017-05-09 21:29:22 -07:00
6e99d36587
Latch error return from nfs_mount_8_cb()
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-09 21:22:46 -07:00
c7995b4fec
Revert "Provide error return from nfs_mount_8_cb()"
...
This reverts commit a4de2ec926
2017-05-09 21:22:06 -07:00
13c253673b
We do not need an async version of readlink2
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-09 14:56:44 -07:00
973d368867
Merge pull request #194 from earlchew/issues-187
...
nfs_readlink2: Return symlink without incurring ENAMETOOLONG
2017-05-09 14:55:27 -07:00
519686356c
Export the nfs_readlink2[_async] symbols from the win32/64 DLL
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-09 14:53:12 -07:00
ebd98bf966
Merge branch 'master' into issues-199
2017-05-09 14:49:58 -07:00
36d5cf8a1c
Merge pull request #204 from earlchew/issues-200
...
Provide error return from nfs_mount_8_cb()
2017-05-09 14:33:03 -07:00
5b28ab5d53
Include sys/sysmacros.h on platforms that have it
...
This addresses issue 202
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-09 14:30:14 -07:00
a4de2ec926
Provide error return from nfs_mount_8_cb()
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-09 07:15:11 -07:00
be6fd2b098
Free inbuf in rpc_destroy_context()
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-09 07:10:23 -07:00
e74846219a
nfs_readlink2: Return symlink without incurring ENAMETOOLONG
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-08 21:56:50 -07:00
e788078e20
Revert "nfs_readlink2: Return symlink without incurring ENAMETOOLONG"
...
This reverts commit 1af36b4c77
2017-05-08 21:55:54 -07:00
ea3ee73e23
Merge pull request #192 from earlchew/issue-191
...
nfs_open: Support O_NOFOLLOW
2017-05-08 21:00:55 -07:00
15c7b4c33a
Add mkdir2() to the symbols we export for win32/64
...
and will at some point restrict export to for .so libraries as well.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-08 20:54:36 -07:00
2589086ce5
Merge pull request #193 from earlchew/issues-189
...
nfs_mkdir2: Include permissions when creating a new directory
2017-05-08 20:52:55 -07:00
3ad4ac1b1e
Merge branch 'master' of github.com:sahlberg/libnfs
2017-05-08 18:40:10 -07:00
b5c5d5901e
Add support for rpc timeouts to the async interface
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-08 18:38:42 -07:00
58ee12fcc7
Merge pull request #198 from earlchew/issues-197
...
Release callback data on nfs_create_2_cb() failure
2017-05-08 15:55:01 -07:00
b356c06eca
Release callback data on nfs_create_2_cb() failure
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-08 15:44:48 -07:00
1af36b4c77
nfs_readlink2: Return symlink without incurring ENAMETOOLONG
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-05 07:57:28 -07:00
adcfda60fa
nfs_mkdir2: Include permissions when creating a new directory
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-05 07:55:09 -07:00
9bd2ceddd3
nfs_open: Support O_NOFOLLOW
...
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
2017-05-05 07:50:35 -07:00
64f7e73ded
Fix bug in nfs_link when using the wrong(invalid) handle when dropping cache
...
This should be a fix for issue 186
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-02 17:50:47 -07:00
24064e96d2
Add example util for nfs_link()
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-02 17:36:27 -07:00
a523f84353
Clamp read/write max size for servers that accept huge PDUs
...
Clamp the max read write size we handle to NFS_MAX_XFER_SIZE for servers
that advertize very large PDU support instead of erroring out.
Fix for issue #188
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-05-01 20:37:35 -07:00
9e2dc9795a
Make nfs_close_async call flush on the file handle before disposing of it.
2017-04-03 14:30:03 -06:00
43feceea8d
don't call rpc_free_pdu() after rpc_queue_pdu() failure
...
rpc_queue_pdu() already calls rpc_free_pdu() in its failure code
path. Doing that again would constitute a double free bug.
2017-03-08 14:29:06 +01:00
70df81c16a
Merge pull request #173 from rosslagerwall/reconnect-retries
...
socket: Limit reconnect retries to 10
2017-03-02 17:07:13 -08:00
7882711c8c
Fix crosscompile with *-w64-mingw32-gcc
2017-02-26 18:14:30 +00:00
08e1cc2f4d
socket: Limit reconnect retries to 10
...
Limit the number of retries when autoreconnecting (to an arbitrary 10)
and return an error to the application if this limit is reached.
Without this, libnfs retries indefinitely and consumes 100% CPU.
See also: https://bugzilla.gnome.org/show_bug.cgi?id=762544
Signed-off-by: Ross Lagerwall <rosslagerwall@gmail.com>
2017-02-26 09:36:09 +00:00
2e0a67fcb9
Merge branch 'master' of github.com:sahlberg/libnfs
2017-02-19 09:15:43 -08:00
6923a99262
Fix memory leak of nested mount list
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-02-19 09:14:29 -08:00
038af3a5ff
socket: pass u_long* to ioctlsocket()
...
See https://msdn.microsoft.com/en-us/library/windows/desktop/ms738573(v=vs.85).aspx
2017-02-10 23:43:21 +01:00
03f3cdf9a9
lib/pdu: remove unnecessary NULL check before free()
...
free(NULL) is explicitly legal.
2017-02-09 12:53:45 +01:00
2b3c164db2
lib/pdu: don't clear pointers in rpc_free_pdu()
...
Pointless, because the containing object is going to be freed.
2017-02-09 12:53:45 +01:00
2f703bd84d
lib/pdu: fix memory leaks in rpc_allocate_*()
...
All error code paths must contain cleanup for all allocations until
that point in the function. Yay for plain C.
2017-02-09 12:52:22 +01:00
906d09f32f
Merge pull request #168 from MaxKellermann/cloexec
...
Use SOCK_CLOEXEC
2017-02-08 17:28:41 -08:00
31cb099f50
init: remove pdu from list before invoking callback
...
If the callback does anything fishy that modifies the linked list,
libnfs may crash after returning. So doing any pending list removals
before invoking the callbacks is safer.
2017-02-08 12:14:52 +01:00
26d90c0663
lib/socket: use SOCK_CLOEXEC in create_socket()
2017-02-08 11:53:00 +01:00
46448bccb4
lib/socket: add function create_socket()
...
Prepare for SOCK_CLOEXEC support.
2017-02-08 11:52:32 +01:00
1d6600652c
Merge pull request #167 from MaxKellermann/static
...
init: move static to beginning of declaration
2017-02-04 17:34:44 -08:00
ddc37a365e
init: move static to beginning of declaration
...
Fixes gcc's -Wold-style-declaration.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
2017-02-03 19:56:35 +01:00
03ac15de5e
Allow passing void* buffers, and make write buffers const
...
Enables callers to pass any opaque data chunk without having to cast
it explicitly.
A write never modifies the source buffer, and thus the pointer should
be const.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
2017-02-03 19:42:12 +01:00
d6a21b2bda
Make rpc_set{g|u}id() public
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
2017-01-30 17:47:18 -08:00
Kevin Vigor