glusterfs.git/xlators/mount, branch v4.0.0-2

fuse: write out reverse notification to fuse dump

2018-01-17T00:00:40+00:00

BUG: 1534602
Change-Id: Ide42cf9cffe462d0cc46272b327c2a05999f09ba
Signed-off-by: Csaba Henk

dict: add more types for values

2018-01-05T09:35:07+00:00

Added 2 more types which are present in gluster codebase, mainly
IATT and UUID.

Updates #203

Change-Id: Ib6d6d6aefb88c3494fbf93dcbe08d9979484968f
Signed-off-by: Amar Tumballi

libglusterfs: fix the call_stack_set_group() function

2017-11-24T16:54:44+00:00

- call_stack_set_group() will take the ownership of passed
  buffer from caller;
- to indicate the change, its signature is changed from
  including the buffer directly to take a pointer to it;
- either the content of the buffer is copied to the
  groups_small embedded buffer of the call stack, or
  the buffer is set as groups_large member of the call
  stack;
- the groups member of the call stack is set to,
  respectively, groups_small or groups_large, according
  to the memory management conventions of the call stack;
- the buffer address is overwritten with junk to effectively
  prevent the caller from using it further on.

Also move call_stack_set_group to stack.c from stack.h
to prevent "defined but not used [-Wunused-function]"
warnings (not using it anymore in call_stack_alloc_group()
implementation, which saved us from this so far).

protocol/server: refactor gid_resolve()

In gid_resolve there are two cases:
either the gid_cache_lookup() call returns
a value or not. The result is caputured in
the agl variable, and throughout the function,
each particular stage of the implementation
comes with an agl and a no-agl variant.

In most cases this is explicitly indicated
via an

   if (agl) {
      ...
   } else {
      ...
   }

but some of this branching are expressed via
goto constructs (obfuscating the fact we stated
above, that is, each particular stage having
an agl/no-agl variant).

In the current refactor, we bring the agl
conditional to the top, and present the
agl/non-agl implementations sequentially.

Also we take the opportunity to clean up and
fix the agl case:
- remove the spurious
    gl.gl_list = agl->gl_list;
  setting, as gl is not used in the agl caae
- populate the group list of call stack from
  agl, fixing thus referred BUG.

Also fixes BUG: 1513920

Change-Id: I61f4574ba21969f7661b9ff0c9dce202b874025d
BUG: 1513928
Signed-off-by: Csaba Henk

mount/fuse: use fstat in getattr implementation if any opened fd is available

2017-11-09T17:26:19+00:00

The restriction of using fds opened by the same Pid means fds cannot
be shared across threads of multithreaded application. Note that fops
from kernel have different Pid for different threads. Imagine
following sequence of operations:

* Turn off performance.open-behind
* Thread t1 opens an fd - fd1 - on file "file". Let's assume nodeid of
  "file" is "nodeid-file".
* Thread t2 does RENAME ("newfile", "file"). Let's assume nodeid of
  "newfile" as "nodeid-newfile".
* t2 proceeds to do fstat (fd1)

The above set of operations can sometimes result in ESTALE/ENOENT
errors. RENAME overwrites "file" with "newfile" changing its nodeid
from "nodeid-file" to "nodeid-newfile" and post RENAME, "nodeid-file" is
removed from the backend. If fstat carries nodeid-file as argument,
which can happen if lookup has not refreshed the nodeid of "file" and
since t2 doesn't have an fd opened, fuse_getattr_resume uses STAT
which will fail as "nodeid-file" no longer exists.

Since the above set of operations and sharing of fds across
multiple threads are valid, this is a bug.

The fix is to use any fd opened on the inode. In this specific example
fuse_getattr_resume will find fd1 and winds down the call as fstat
(fd1) which won't fail.

Cross-checked with "Miklos Szeredi"  for
any security issues with this solution and he approves the solution.

Thanks to "Miklos Szeredi"  for all the
pointers and discussions.

Change-Id: I88dd29b3607cd2594eee9d72a1637b5346c8d49c
BUG: 1510401
Signed-off-by: Raghavendra G

core: make gf_boolean_t a C99 bool instead of an enum

2017-11-03T05:04:11+00:00

This reduces the space used from four bytes to one, and allows
new code to use familiar C99 types/values interoperably with our
old cruft. It does *not* change current declarations or code;
that will be left for a separate - much larger - patch.

Updates: #80

Change-Id: I5baedd17d3fb05b38f0d8b8bb9dd62824475842e
Signed-off-by: Jeff Darcy

gfproxyd: Let glusterd manage gfproxy daemon

2017-10-18T14:15:05+00:00

Updates: #242
BUG: 1428063
Change-Id: Iaaf2edf99b2ecc75f6d30762c752a6d445c1c826
Signed-off-by: Poornima G

mount/fuse: never fail open(dir) with ENOENT

2017-10-17T05:15:32+00:00

open(dir) being an operation on inode should never fail with
ENOENT. If gfid is not present, the appropriate error is ESTALE. This
will enable kernel to retry open after a revalidate lookup.

Change-Id: I8d07d2ebb5a0da6c3ea478317442cb42f1797a4b
BUG: 1500269
Signed-off-by: Raghavendra G

Revert "mount/fuse: report ESTALE as ENOENT"

2017-10-17T05:14:41+00:00

This reverts commit 26d16b90ec7f8acbe07e56e8fe1baf9c9fa1519e.

Consider rename (index.new, store.idx) and open (store.idx) being
executed in parallel. When we break down operations following sequence
is possible.

* lookup (store.idx) - as part of open(store.idx) returns gfid1 as the
  result.
* rename (index.new, store.idx) changes gfid of store.idx to
  gfid2. Note that gfid2 was the nodeid of index.new. Since rename is
  successful, gfid2 is associated with store.idx.
* open (store.idx) resumes and issues open fop to glusterfs with
  gfid1. open in glusterfs fails as gfid1 doesn't exist and the error
  returned by glusterfs to kernel-fuse is ENOENT.
* kernel passes back the same error to application as a result to
  open.

This error could've been prevented if kernel retries open with
gfid2. Interestingly kernel do retry open when it receives ESTALE
error. Even though failure to find gfid resulted in ESTALE error,
commit 26d16b90ec7f8acb converted that error to ENOENT while sending
an error reply to kernel. This prevented kernel from retrying open
resulting in error.

Change-Id: I2e752ca60dd8af1b989dd1d29c7b002ee58440b4
BUG: 1500269
Signed-off-by: Raghavendra G

mount/fuse : Fix parsing of vol_id for snapshot volume

2017-10-15T14:43:02+00:00

For supporting sub-dir mount, we changed the volid. Which means anything
after a '/' in volume_id will be considered as sub-dir path.

But snapshot volume has vol_id stracture of /snaps//
which has to be considered as during the parsing.

Note 1: sub-dir mount is not supported on snapshot volume
Note 2: With sub-dir mount changes brick based mount for quota cannot be
        executed via mount command. It has to be a direct call via glusterfs

Change-Id: I0d824de0236b803db8a918f683dabb0cb523cb04
BUG: 1501235
Signed-off-by: Mohammed Rafi KC

mount/fuse: Make event-history feature configurable

2017-09-24T06:22:07+00:00

... and disable it by default.

This is because having it disabled seems to improve performance.
This could be due to the lock contention by the different epoll threads
on the circular buff lock in the fop cbks just before writing their response
to /dev/fuse.

Just to provide some data - wrt ovirt-gluster hyperconverged
environment, I saw an increase in IOPs by 12K with event-history
disabled for randrom read workload.

Usage:
mount -t glusterfs -o event-history=on $HOSTNAME:$VOLNAME $MOUNTPOINT
OR
glusterfs --event-history=on --volfile-server=$HOSTNAME --volfile-id=$VOLNAME $MOUNTPOINT

Change-Id: Ia533788d309c78688a315dc8cd04d30fad9e9485
BUG: 1467614
Signed-off-by: Krutika Dhananjay