Commit graph

55 commits

Author SHA1 Message Date
Pierre Schweitzer
901c47ed14
[FASTFAT] Don't delay any other close once shutdown has started 2018-08-19 09:56:12 +02:00
Pierre Schweitzer
4fe7aafe10
[FASTFAT] Magic values-- 2018-08-19 09:56:12 +02:00
Pierre Schweitzer
047479e0a2
[FASTFAT] Don't leak statistics on dismount 2018-08-18 19:03:30 +02:00
Pierre Schweitzer
639e6d635f
[FASTFAT] Reduce the usage of the generic allocation tag
And use tag compatible with MS FastFAT, to use debug with WinDBG
2018-08-18 19:03:30 +02:00
Pierre Schweitzer
50b00f0fcc
[FASTFAT] Implement delayed close
When we're about to close a file (ie, forget everything about it
and release any associated structure), actually delay it.
This allows keep data fresh in memory for faster reuse in case
it would be required. The effective closing will only happen after some time.

For specific operations, this will produce a real speed up in ReactOS.
For instance, with that patch, Winamp starts within seconds, instead of dozen
of minutes.
In most cases, it will bring ReactOS to performances it had before fixing
the huge leak in FastFAT (commit 94ead99) without leaking the whole FS.

For now, due to regressions, this is only activated for files and not
for directories. Once it gets fixed, it will be enabled for both.

CORE-14826
CORE-14917
2018-08-18 19:03:30 +02:00
Pierre Schweitzer
52f0f80a83
[FASTFAT] Don't add an extra \, when renaming a file at root.
This fixes failures to rename a file where destination is
the root of a FAT volume.

CORE-10503
2018-06-09 22:17:17 +02:00
Pierre Schweitzer
6aa4beeefb
[FASTFAT] Use the FastFAT mechanism for counting clusters already implemented
This allows us having more accurate statistics regarding available clusters
count. Even though FastFAT and chkdsk still don't agree!

CORE-3877
2018-06-09 18:23:07 +02:00
Pierre Schweitzer
704baed2df
[FASTFAT] Start implementing FAT32 statistics support
Not fully perfect yet, chkdsk still complains.

CORE-3877
2018-06-09 12:36:43 +02:00
Pierre Schweitzer
04640f9264
[FASTFAT] Save the FSINFO sector location 2018-06-09 12:11:43 +02:00
Pierre Schweitzer
01e2877161
[FASTFAT] Properly check for buffer size on dir enumeration.
This avoids improper returns when multiple entries are requested
and, in rare circumstances, entries not being seen by umode.

CORE-14699
2018-06-07 19:34:47 +02:00
Pierre Schweitzer
9f3169b295
[FASTFAT] Lock DirResource on flush in case we're updating entry
CORE-14660
2018-05-28 19:37:56 +02:00
Pierre Schweitzer
fc788cf2fd
[FASTFAT] Lock DirResource when modifying an entry on disk.
Likely not optimal, but fixes some races conditions where
the directory is uninit in the middle of the write.
2018-05-26 17:13:08 +02:00
Pierre Schweitzer
47b9db9d02
[FASTFAT] Initialize cache if neded for changing volume label.
CORE-14654
2018-05-26 10:12:55 +02:00
Pierre Schweitzer
b4363068d1
[FASTFAT] Properly handle IRPs that can wait and these that cannot.
CORE-14634
2018-05-22 21:30:08 +02:00
Pierre Schweitzer
e9e14be661
[FASTFAT] Properly queue pending write IRPs.
Addendum to 30b836b
2018-05-21 10:41:59 +02:00
Pierre Schweitzer
30b836bf3f
[FASTFAT] When extending a file, also lock the DirResource.
This avoids race conditions under high IOs and thus corruption on
the FS, or assertions failures in the kernel.
Easily triggered by building ReactOS on ReactOS ;-).
2018-05-21 10:30:14 +02:00
Pierre Schweitzer
a10f6c7a0c
[FASTFAT] Fail if allocating the stream FO fails. 2018-05-20 13:53:43 +02:00
Pierre Schweitzer
7283f3cfd9
[FASTFAT] Addendum to d69f318 2018-05-19 13:46:34 +02:00
Pierre Schweitzer
d69f31848e
[FASTFAT] Misc. fixes spotted by Thomas. 2018-05-19 10:07:14 +02:00
Pierre Schweitzer
7c01587680
[FASTFAT] Completely rewrite support for dirty volumes.
Until now, our support for dirty volumes was totally broken
to a point where, on FAT32 volume, the dirty couldn't even
be written nor read from the disk.

This commit totally rewrites its handling, for both FAT16 and FAT32
so that it's now fully functionnal. Furthermore, it also gets
totally compatible with our vfatlib, and thus, autochk.
Now, on mount, FastFAT will check if the volume is dirty or not, and
autochk will be able to ask for a repair if dirty. vfatlib will
repair the volume and remove the dirty bit. So that, on next
reboot, the volume will be mounted clean.

As a reminder, the dirty bit is set immediately after mounting
the volume, so that, if you crash or have a powercut, autochk
will always attempt to repair your volume (with more or less,
that's FAT!).

If you want to experience without breaking your FAT volume,
just boot, open a cmd prompt and type: fsutil dirty set c:
and reboot!

CORE-13758
CORE-13760
CORE-13759
2018-05-18 23:05:05 +02:00
Pierre Schweitzer
efa75dd5b2
[FASTFAT] Allow locking system volume on boot.
This is a hack, and totally not the default behavior.
But it will allow autochk locking the boot volume in
order to verify it on boot, in case it would have
been improperly dismounted.

CORE-13759
2018-05-18 23:05:05 +02:00
Pierre Schweitzer
a837694518
[FASTFAT] Uninit directory cache on last handle close.
CORE-14629
CORE-14557
2018-05-18 14:51:24 +02:00
Pierre Schweitzer
3c3ebe3320
[FASTFAT] Only initialize directory cache on use.
This avoids initializing cache directly on directory
open/create.
The advantage is we reduce the load on cache manager
and on memory manager by avoiding creating everytime
a stream file object, and initializing cache for it.

This will avoid initializing cache for started
applications 'current directory' which is just opened
for having a valid handle but no read/write is performed
in it, by default.

This is a step forward for autochk.

CORE-14629
2018-05-18 14:09:30 +02:00
Pierre Schweitzer
5d743b7bd7
[FASTFAT] When attaching our FCB to a FO, also set the VPB.
Spotted by Alex.
2018-05-18 10:30:52 +02:00
Pierre Schweitzer
d37280efaa
[FASTFAT] Deny dismounting system volume or a volume with a pagefile 2018-05-16 21:45:35 +02:00
Pierre Schweitzer
38078335b9
[FASTFAT] Implement support for stack overflow in read operations.
Before performing a read operation, FastFAT driver will
attempt to compute whether it would run out of stack
during the operation. If so, instead of attempting the
operation in the current thread, it will post the read
request to the overflow thread.

This should help with the regressions brought in by
94ead99e0c.

CORE-14601
2018-05-12 12:03:18 +02:00
Pierre Schweitzer
2a7d16727a
[FASTFAT] Implement write IOs defering.
Before any write operation that would involve caching, ask
the cache controler whether writing would make it exceed its memory
consumption. If so, queue the write operation for later execution.

In case the write operation can wait, then, the FSD operation will be
halted until the write is allowed.

I could test it successfully by copying huge files from a FAT volume to
another. The write is halted until some portions of the file is written
to the disk.
I could also properly install Qt (SDK) on ReactOS with this and less than 1GB RAM:
- https://www.heisspiter.net/~Pierre/rostests/Qt_OS.png
- https://www.heisspiter.net/~Pierre/rostests/Qt_OS2.png

CORE-12081
CORE-14582
CORE-14313
2018-04-29 20:42:53 +02:00
Pierre Schweitzer
00e843b1ed
[FASTFAT] Reduce the number of local declaration. 2018-04-29 12:33:53 +02:00
Pierre Schweitzer
fd8197645f
[FASTFAT] Use KdRosRegisterCliCallback() to register in KDBG. 2018-04-29 12:23:18 +02:00
Pierre Schweitzer
c5a35ecd37 [FASTFAT] Introduce a KDBG extension.
This is a PoC of what it's possible to realize thanks to an
already existing hack in ntoskrnl :-).
With this extension, on the kdb:> prompt, you're able to type
in commands that will be handled by the FastFAT driver and not
by the kernel, allowing internal debug, not possible otherwise.

So far, three commands exist:
- ?fat.vols: lists all the mounted volumes by FastFAT
- ?fat.files: lists all the files on a specific volume (with their attributes)
- ?fat.setdbgfile: allows watching on specifics files lifetime

This is obviously only the begin and could be greatly improved.

For instance, this is what allowed to debug CORE-14557
2018-04-29 12:15:11 +02:00
Pierre Schweitzer
94ead99e0c [FASTFAT] Don't leak directories FILE_OBJECT, FCB and cache entries.
Once a directory is crossed (opened or a child is opened), associated
FCB structure is created in FastFAT, but also a stream FO for caching.
Up to now, due to an extra reference taken by the stream file object,
even when the directory was no longer used, the directory was kept in
memory: the FCB was never deleted, the file object was never dereferenced,
and the cache never released.

The immediate effect of this bug is that our FAT driver was leaking every
directory that was used affecting the whole OS situation. In case of
directories intensive operation (like extraction the ReactOS source code
in ReactOS ;-)), we were just killin the whole OS RAM without any way to
release it and recover.

The other side effects: IOs were faster as half of the FS was always
permanant in RAM.

This commit fixes the issue by forcing the FSD to release the FO,
and the cache when a directory is no longer used, leading to its
destruction in RAM.
Downside: on IO intensive operation, expect slowdowns, obviously,
there's less caching now. But more efficient!

CORE-14557
2018-04-28 18:33:14 +02:00
Pierre Schweitzer
315f9a13d4
[FASTFAT] When allocating a FCB, set its node type code.
We set the same value as MS FastFAT driver for consistency.
2018-04-27 18:56:31 +02:00
Pierre Schweitzer
69f51d1533
[FASTFAT] Drop read ahead locking routines in favor of lazy writer routines.
This avoids code duplication while being consistent with MS FastFAT.
2018-02-17 13:49:15 +01:00
Serge Gautherie
d6fab4158e [FASTFAT] Improve code style in addition to 8294118174. 2018-01-07 19:29:23 +01:00
Pierre Schweitzer
f0a1a5c773
[FASTFAT] Notify on file size change in FileAllocationInformation/FileEndOfFileInformation 2018-01-07 15:29:33 +01:00
Pierre Schweitzer
8294118174
[FASTFAT] Add a wrapper around FsRtlNotifyFullReportChange 2018-01-07 14:16:31 +01:00
Pierre Schweitzer
2a36b632ef
[FASTFAT] When a file gets overwritten, notify about the changes 2018-01-07 11:44:25 +01:00
Pierre Schweitzer
ffeb3f070b
[FASTFAT] When overwriting a file, also update its write time 2018-01-07 11:34:41 +01:00
Pierre Schweitzer
0eacb0b1ef
[FASTFAT] When dealing with attributes on IRP_MJ_CREATE, filter user input to ensure we have only attributes we can handle.
This applies to files newly created, or for files overwritten where attributes are reset/modified.
2018-01-06 22:02:10 +01:00
Pierre Schweitzer
2b4d5c5cff
[FASTFAT] When overwriting a file with FILE_OVERWRITE*, don't erase previous file attributes, only update them.
Keep erasing them in case the file gets overwritten with FILE_SUPERSEDE.

CORE-14158
2018-01-01 14:29:15 +01:00
Pierre Schweitzer
f3fecf92c1
[FASTFAT] On file overwrite, deny it if file as HIDDEN/SYSTEM flag and caller doesn't set them.
CORE-14158
2018-01-01 12:13:46 +01:00
Pierre Schweitzer
453e42beb5
[FASTFAT] Revert bd1e7bf: our VPB swapout implementation in FastFAT is buggy and
leaves volumes in half-(dis)mounted state and thus they get remounted while not
completely dismounted.
This can lead to major breakage and data corruption.
This requires deeper fixes (let's just drop that driver!).

CORE-14124
CORE-14126
CORE-14133
2017-12-25 11:24:13 +01:00
Pierre Schweitzer
bd1e7bf85e
[FASFAT] Enable FAT volume dismount using VPB swapout 2017-12-17 23:29:33 +01:00
Pierre Schweitzer
dcd95c1a85
[FASTFAT] Add a comment (and an ASSERT) in VfatCreateFile() that shows that my stats code is broken... 2017-12-17 18:25:43 +01:00
Pierre Schweitzer
64bc96558e
[FASTFAT] While closing FCBs on dismount, release from tail to head and not the contrary.
It fixes assertion failure in vfatDestroyFCB() where we would have release parent before child.
This is still not perfect, but less bug prone...

With this commits (and ENABLE_SWAPOUT defined), ReactOS seems to unmount FAT volumes quite nice! :-)
(Tried with fsutil volume dismount X:)
2017-12-17 18:24:01 +01:00
Pierre Schweitzer
de03686148
[FASTFAT] Fix a handle count leak on volume close. This can prevent locking a volume! 2017-12-17 18:24:01 +01:00
Pierre Schweitzer
27773dbb3f
[FASTFAT] When not able to lock a volume due to open handles, print open handles 2017-12-17 18:24:01 +01:00
Pierre Schweitzer
18d6584da4
[FASTFAT] Fix FastFAT not returning short name for FAT volumes in FileBothDirectoryInformation case
This is likely due to a copy paste error where long name was copied twice and short never.
Fun fact: this was not affecting FATX volumes
Fun fact2: this was defeating a buffer overflow check and thus was allowing buffer overflow!

CORE-14088
2017-12-09 21:22:55 +01:00
Pierre Schweitzer
bddb0db469
[FASTFAT] Finally drop the FileNameInformation class for directories.
It makes no sense for them and shouldn't be implemented.
It's an addendum to 9f3c801.
2017-12-09 14:37:07 +01:00
Pierre Schweitzer
28343c6c0c
[FASTFAT] Fix size checking in VfatGetFileNameInformation() 2017-12-09 12:45:16 +01:00