Commit Graph

433 Commits

Author SHA1 Message Date
Ian Lepore
74938cbb7f Make the F_READAHEAD option to fcntl(2) work as documented: a value of zero
now disables read-ahead.  It used to effectively restore the system default
readahead hueristic if it had been changed; a negative value now restores
the default.

Reviewed by:	kib
2013-02-13 15:09:16 +00:00
Pawel Jakub Dawidek
a2c496ebb9 Remove label that was accidentally moved during Giant removal from VFS. 2013-01-31 22:14:16 +00:00
Dag-Erling Smørgrav
b5471c918f Rewrite fdgrowtable() so common mortals can actually understand what
it does and how, and add comments describing the data structures and
explaining how they are managed.
2012-12-20 20:18:27 +00:00
Konstantin Belousov
5050aa86cf Remove the support for using non-mpsafe filesystem modules.
In particular, do not lock Giant conditionally when calling into the
filesystem module, remove the VFS_LOCK_GIANT() and related
macros. Stop handling buffers belonging to non-mpsafe filesystems.

The VFS_VERSION is bumped to indicate the interface change which does
not result in the interface signatures changes.

Conducted and reviewed by:	attilio
Tested by:	pho
2012-10-22 17:50:54 +00:00
Konstantin Belousov
d8c1da8b90 Add F_DUP2FD_CLOEXEC. Apparently Solaris 11 already did this.
Submitted by:	Jukka A. Ukkonen <jau iki fi>
PR:	standards/169962
MFC after:	1 week
2012-07-27 10:41:10 +00:00
Konstantin Belousov
a53cab2c6c (Incomplete) fixes for symbols visibility issues and style in fcntl.h.
Append '__' prefix to the tag of struct oflock, and put it under BSD
namespace. Structure is needed both by libc and kernel, thus cannot be
hidden under #ifdef _KERNEL.

Move a set of non-standard F_* and O_* constants into BSD namespace.
SUSv4 explicitely allows implemenation to pollute F_* and O_* names
after fcntl.h is included, but it costs us nothing to adhere
to the specification if exact POSIX compliance level is requested by
user code.

Change some spaces after #define to tabs.

Noted by and discussed with:	     bde
MFC after:   1 week
2012-07-21 13:02:11 +00:00
Konstantin Belousov
eb3d975443 Remove line which was accidentally kept in r238614.
Submitted by:	pjd
Pointy hat to:	kib
MFC after:	1 week
2012-07-19 20:38:03 +00:00
Konstantin Belousov
49d02b13bc Implement F_DUPFD_CLOEXEC command for fcntl(2), specified by SUSv4.
PR:	  standards/169962
Submitted by:	Jukka A. Ukkonen <jau iki fi>
MFC after:	1 week
2012-07-19 10:22:54 +00:00
Mateusz Guzik
4fd85c4b5d Follow-up commit to r238220:
Pass only FEXEC (instead of FREAD|FEXEC) in fgetvp_exec. _fget has to check for
!FWRITE anyway and may as well know about FREAD.

Make _fget code a bit more readable by converting permission checking from if()
to switch(). Assert that correct permission flags are passed.

In collaboration with:	kib
Approved by:	trasz (mentor)
MFC after:	6 days
X-MFC: with r238220
2012-07-09 05:39:31 +00:00
Mateusz Guzik
28a7f60741 Unbreak handling of descriptors opened with O_EXEC by fexecve(2).
While here return EBADF for descriptors opened for writing (previously it was ETXTBSY).

Add fgetvp_exec function which performs appropriate checks.

PR:		kern/169651
In collaboration with:	kib
Approved by:	trasz (mentor)
MFC after:	1 week
2012-07-08 00:51:38 +00:00
Konstantin Belousov
c5c1199c83 Extend the KPI to lock and unlock f_offset member of struct file. It
now fully encapsulates all accesses to f_offset, and extends f_offset
locking to other consumers that need it, in particular, to lseek() and
variants of getdirentries().

Ensure that on 32bit architectures f_offset, which is 64bit quantity,
always read and written under the mtxpool protection. This fixes
apparently easy to trigger race when parallel lseek()s or lseek() and
read/write could destroy file offset.

The already broken ABI emulations, including iBCS and SysV, are not
converted (yet).

Tested by:	pho
No objections from:	jhb
MFC after:    3 weeks
2012-07-02 21:01:03 +00:00
Pawel Jakub Dawidek
d99e1d5fd6 Don't check for race with close on advisory unlock (there is nothing smart we
can do when such a race occurs). This saves lock/unlock cycle for the filedesc
lock for every advisory unlock operation.

MFC after:	1 month
2012-06-17 21:04:22 +00:00
Pawel Jakub Dawidek
604a7c2f00 Extend the comment about checking for a race with close to explain why
it is done and why we don't return an error in such case.

Discussed with:	kib
MFC after:	1 month
2012-06-17 16:59:37 +00:00
Pawel Jakub Dawidek
fd6049b186 If VOP_ADVLOCK() call or earlier checks failed don't check for a race with
close, because even if we had a race there is nothing to unlock.

Discussed with:	kib
MFC after:	1 month
2012-06-17 16:32:32 +00:00
Pawel Jakub Dawidek
cff2dcd10d Revert r237073. 'td' can be NULL here.
MFC after:	1 month
2012-06-16 12:56:36 +00:00
Pawel Jakub Dawidek
3cde71cb25 One more attempt to make prototypes formated according to style(9), which
holefully recovers from the "worse than useless" state.

Reported by:	bde
MFC after:	1 month
2012-06-15 10:00:29 +00:00
Pawel Jakub Dawidek
19a8f6748e Remove fdtofp() function and use fget_locked(), which works exactly the same.
MFC after:	1 month
2012-06-14 16:25:10 +00:00
Pawel Jakub Dawidek
b7fc69ca89 Assert that the filedesc lock is being held when the fdunwrap() function
is called.

MFC after:	1 month
2012-06-14 16:23:16 +00:00
Pawel Jakub Dawidek
1a94dc8581 Simplify the code by making more use of the fdtofp() function.
MFC after:	1 month
2012-06-14 15:37:15 +00:00
Pawel Jakub Dawidek
215aeba939 - Assert that the filedesc lock is being held when fdisused() is called.
- Fix white spaces.

MFC after:	1 month
2012-06-14 15:35:14 +00:00
Pawel Jakub Dawidek
7aef754274 Style fixes and assertions improvements.
MFC after:	1 month
2012-06-14 15:34:10 +00:00
Pawel Jakub Dawidek
8d169d9ff0 Assert that the filedesc lock is not held when closef() is called.
MFC after:	1 month
2012-06-14 15:26:23 +00:00
Pawel Jakub Dawidek
eb273c01f3 Style fixes.
Reported by:	bde
MFC after:	1 month
2012-06-14 15:21:57 +00:00
Pawel Jakub Dawidek
c7e9a659ca Remove code duplication from fdclosexec(), which was the reason of the bug
fixed in r237065.

MFC after:	1 month
2012-06-14 12:43:37 +00:00
Pawel Jakub Dawidek
8f59e9fddc When we are closing capabilities during exec, we want to call mq_fdclose()
on the underlying object and not on the capability itself.

Similar bug was fixed in r236853.

MFC after:	1 month
2012-06-14 12:41:21 +00:00
Pawel Jakub Dawidek
5570ae7d87 Style.
MFC after:	1 month
2012-06-14 12:37:41 +00:00
Pawel Jakub Dawidek
620216725a When checking if file descriptor number is valid, explicitely check for 'fd'
being less than 0 instead of using cast-to-unsigned hack.

Today's commit was brought to you by the letters 'B', 'D' and 'E' :)
2012-06-13 22:12:10 +00:00
Pawel Jakub Dawidek
3812dcd3de Allocate descriptor number in dupfdopen() itself instead of depending on
the caller using finstall().
This saves us the filedesc lock/unlock cycle, fhold()/fdrop() cycle and closes
a race between finstall() and dupfdopen().

MFC after:	1 month
2012-06-13 21:32:35 +00:00
Pawel Jakub Dawidek
6195bfebcc There is only one caller of the dupfdopen() function, so we can simplify
it a bit:
- We can assert that only ENODEV and ENXIO errors are passed instead of
  handling other errors.
- The caller always call finstall() for indx descriptor, so we can assume
  it is set. Actually the filedesc lock is dropped between finstall() and
  dupfdopen(), so there is a window there for another thread to close the
  indx descriptor, but it will be closed in next commit.

Reviewed by:	mjg
MFC after:	1 month
2012-06-13 19:00:29 +00:00
Mateusz Guzik
2ca63f0a90 Remove 'low' argument from fd_last_used().
This function is static and the only caller always passes 0 as low.

While here update note about return values in comment.

Reviewed by:	pjd
Approved by:	trasz (mentor)
MFC after:	1 month
2012-06-13 17:18:16 +00:00
Mateusz Guzik
02efb9a8b1 Re-apply reverted parts of r236935 by pjd with some changes.
If fdalloc() decides to grow fdtable it does it once and at most doubles
the size. This still may be not enough for sufficiently large fd. Use fd
in calculations of new size in order to fix this.

When growing the table, fd is already equal to first free descriptor >= minfd,
also fdgrowtable() no longer drops the filedesc lock. As a result of this there
is no need to retry allocation nor lookup.

Fix description of fd_first_free to note all return values.

In co-operation with:	pjd
Approved by:	trasz (mentor)
MFC after:	1 month
2012-06-13 17:12:53 +00:00
Pawel Jakub Dawidek
faf0db351d Revert part of the r236935 for now, until I figure out why it doesn't
work properly.

Reported by:	davidxu
2012-06-12 10:25:11 +00:00
Pawel Jakub Dawidek
039dc89f0d fdgrowtable() no longer drops the filedesc lock so it is enough to
retry finding free file descriptor only once after fdgrowtable().

Spotted by:	pluknet
MFC after:	1 month
2012-06-11 22:05:26 +00:00
Pawel Jakub Dawidek
d3ec30e525 Use consistent way of checking if descriptor number is valid.
MFC after:	1 month
2012-06-11 20:17:20 +00:00
Pawel Jakub Dawidek
fd45a47ba6 Be consistent with white spaces.
MFC after:	1 month
2012-06-11 20:01:50 +00:00
Pawel Jakub Dawidek
19d9c0e11e Remove code duplicated in kern_close() and do_dup() and use closefp() function
introduced a minute ago.

This code duplication was responsible for the bug fixed in r236853.

Discussed with:	kib
Tested by:	pho
MFC after:	1 month
2012-06-11 20:00:44 +00:00
Pawel Jakub Dawidek
642db963ab Introduce closefp() function that we will be able to use to eliminate
code duplication in kern_close() and do_dup().

This is committed separately from the actual removal of the duplicated
code, as the combined diff was very hard to read.

Discussed with:	kib
Tested by:	pho
MFC after:	1 month
2012-06-11 19:57:31 +00:00
Pawel Jakub Dawidek
129c87eb7d Merge two ifs into one to make the code almost identical to the code in
kern_close().

Discussed with:	kib
Tested by:	pho
MFC after:	1 month
2012-06-11 19:53:41 +00:00
Pawel Jakub Dawidek
d327cee241 Move the code around a bit to move two parts of code duplicated from
kern_close() close together.

Discussed with:	kib
Tested by:	pho
MFC after:	1 month
2012-06-11 19:51:27 +00:00
Pawel Jakub Dawidek
8b40793150 Now that fdgrowtable() doesn't drop the filedesc lock we don't need to
check if descriptor changed from under us. Replace the check with an
assert.

Discussed with:	kib
Tested by:	pho
MFC after:	1 month
2012-06-11 19:48:55 +00:00
Pawel Jakub Dawidek
69d7614850 When we are closing capability during dup2(), we want to call mq_fdclose()
on the underlying object and not on the capability itself.

Discussed with:	rwatson
Sponsored by:	FreeBSD Foundation
MFC after:	1 month
2012-06-10 14:57:18 +00:00
Pawel Jakub Dawidek
1b693d7494 Merge two ifs into one. Other minor style fixes.
MFC after:	1 month
2012-06-10 13:10:21 +00:00
Pawel Jakub Dawidek
8849ae7256 Simplify fdtofp().
MFC after:	1 month
2012-06-10 06:31:54 +00:00
Pawel Jakub Dawidek
e59a97362d There is no need to drop the FILEDESC lock around malloc(M_WAITOK) anymore, as
we now use sx lock for filedesc structure protection.

Reviewed by:	kib
MFC after:	1 month
2012-06-09 18:50:32 +00:00
Pawel Jakub Dawidek
68abac4337 Remove now unused variable.
MFC after:	1 month
MFC with:	r236820
2012-06-09 18:48:06 +00:00
Pawel Jakub Dawidek
380513aaae Make some of the loops more readable.
Reviewed by:	tegge
MFC after:	1 month
2012-06-09 18:03:23 +00:00
Pawel Jakub Dawidek
5d02ed91e9 Correct panic message.
MFC after:	1 month
MFC with:	r236731
2012-06-09 12:27:30 +00:00
Pawel Jakub Dawidek
bf3e37ef15 In fdalloc() f_ofileflags for the newly allocated descriptor has to be 0.
Assert that instead of setting it to 0.

Sponsored by:	FreeBSD Foundation
MFC after:	1 month
2012-06-07 23:33:10 +00:00
Eitan Adler
847d0034e3 Return EBADF instead of EMFILE from dup2 when the second argument is
outside the range of valid file descriptors

PR:		kern/164970
Submitted by:	Peter Jeremy <peterjeremy@acm.org>
Reviewed by:	jilles
Approved by:	cperciva
MFC after:	1 week
2012-04-11 14:08:09 +00:00
John Baldwin
e506e182dd Export some more useful info about shared memory objects to userland
via procstat(1) and fstat(1):
- Change shm file descriptors to track the pathname they are associated
  with and add a shm_path() method to copy the path out to a caller-supplied
  buffer.
- Use the fo_stat() method of shared memory objects and shm_path() to
  export the path, mode, and size of a shared memory object via
  struct kinfo_file.
- Add a struct shmstat to the libprocstat(3) interface along with a
  procstat_get_shm_info() to export the mode and size of a shared memory
  object.
- Change procstat to always print out the path for a given object if it
  is valid.
- Teach fstat about shared memory objects and to display their path,
  mode, and size.

MFC after:	2 weeks
2012-04-01 18:22:48 +00:00