Commit graph

174 commits

Author SHA1 Message Date
Jérôme Gardou 2a962eaf8c [NTOS:CC] Keep a reference on the shared cache map of the file when we are in lazy write
This should fix "Leaking VACB" debug prints
2021-02-19 15:48:31 +01:00
Timo Kreuzer ca74467904
[NTOS:CC] Raise status, when CcMapData fails (#3348)
Also add a check in fastfat, that asserts, when CcMapData returns FALSE.
2021-02-09 09:29:48 +01:00
Jérôme Gardou cc9607e94e [NTOS:CC] Fix use of uninitialized variable 2021-02-03 13:35:17 +01:00
Jérôme Gardou b7eb0fddf3 Address PR review 2021-02-03 09:41:24 +01:00
Jérôme Gardou 2ba1926037 [NTOS:MM][NTOS:CC] Performance improvement again
Read files by 64kb chunks instead of page-sized chunks.
2021-02-03 09:41:23 +01:00
Jérôme Gardou 7fbf90d6fd [NTOS:CC] Fix a bug 2021-02-03 09:41:23 +01:00
Jérôme Gardou cc6bc02b91 [NTOS:CC] Make sure to purge the Mm side in CcPurgeCacheSection 2021-02-03 09:41:23 +01:00
Jérôme Gardou d0bf98663b [NTOS:CC] Be sure to flush the whole file in CcFlushCache 2021-02-03 09:41:23 +01:00
Jérôme Gardou 5949d5095d [NTOS:CC][NTOS:MM] Try respecting ValidDataLength 2021-02-03 09:41:23 +01:00
Jérôme Gardou bdb73edab7 [NTOS:CC] Flush the whole VACB
Let Mm know what it has to do.
2021-02-03 09:41:23 +01:00
Jérôme Gardou 804f5a41ed [NTOS:CC] Improve trace messages 2021-02-03 09:41:23 +01:00
Jérôme Gardou 20fe42c9e9 [NTOS:CC] Simplify CcFlushCache implementation 2021-02-03 09:41:23 +01:00
Jérôme Gardou 9b6240ee03 [NTOS:CC] Get rid of ROS_VACB:Valid 2021-02-03 09:41:23 +01:00
Jérôme Gardou 02167f2fb3 [NTOS:CC] Properly return the public BCBs 2021-02-03 09:41:23 +01:00
Jérôme Gardou 33cde28312 [NTOS:CC] Simplify CcRosDeleteFileCache 2021-02-03 09:41:23 +01:00
Jérôme Gardou a9193b5cc2 [NTOS:CC] Remove dead code 2021-02-03 09:41:23 +01:00
Jérôme Gardou 57ee31ee33 [NTOS:CC] Perform sanity checks before doing anything else 2021-02-03 09:41:23 +01:00
Jérôme Gardou 8287a098b9 [NTOS:CC] Fix potnetial use-after-free 2021-02-03 09:41:23 +01:00
Jérôme Gardou 358d06c47c [NTOS:CC] Always honor WriteThrough parameter in CcUnpinRepinnedBcb 2021-02-03 09:41:23 +01:00
Jérôme Gardou 1505abbc09 [NTOS:CC] Do not write behind concurrently the same file 2021-02-03 09:41:23 +01:00
Jérôme Gardou 347a4f146b [NTOS] Loop again and again until the whole cache is empty when sutting down 2021-02-03 09:41:23 +01:00
Jérôme Gardou 91f6fa0259 Disable CcScheduleReadAhead call 2021-02-03 09:41:22 +01:00
Jérôme Gardou 6d97d8d2e1 [NTOS:CC] Fix some tests, complain where the current implementation won't let us do the right thing 2021-02-03 09:41:22 +01:00
Jérôme Gardou 30f71c7fc0 [NTOS] Zero data unconditionally after segment end, unless section is created with SEC_RESERVE
Use a SEC_RESERVE section in Cc
2021-02-03 09:41:22 +01:00
Jérôme Gardou a52bc6d179 [NTOS:CC] Restore read-ahead behaviour in CcCopyRead 2021-02-03 09:41:22 +01:00
Jérôme Gardou 4441b1cd44 [NTOS:CC] Fix some tests for CcCopyRead and CcCopyWrite
Most importantly: raise the right status when provided an invalid buffer.
2021-02-03 09:41:22 +01:00
Jérôme Gardou cd23d6e19a [NTOS:CC] Honor FO_WRITE_THROUGH flag in CcCopyWrite and fix it in CcZeroData 2021-02-03 09:41:22 +01:00
Jérôme Gardou f8aa14ce4e [NTOS:CC] Acquire file for flush when flushing if necessary 2021-02-03 09:41:22 +01:00
Jérôme Gardou cbfc04157d [NTOS:CC] File size check fix 2021-02-03 09:41:22 +01:00
Jérôme Gardou 70c62aa2c9 [NTOS:CC] Fix Vacb size usage & check 2021-02-03 09:41:22 +01:00
Jérôme Gardou c74cbf0c0b [NTOS/CC] Be more precise when notifying Mm about dirty pages 2021-02-03 09:41:22 +01:00
Jérôme Gardou c295d6b229 [NTOS:CC] Fix releasing VACB when an exception is raised in CcZeroData & CcCopyWrite 2021-02-03 09:41:22 +01:00
Jérôme Gardou 7f47f03296 [NTOS:CC] Rewrite the non-cached path of CcZeroData 2021-02-03 09:41:22 +01:00
Jérôme Gardou 36e18aab35 [NTOS:CC] Remove unused functions 2021-02-03 09:41:22 +01:00
Jérôme Gardou d8cdb89fb0 [NTOSKRNL] Overhaul Cc and Mm relationship
Previously, when creating a file section, Mm requested Cc to cache the file, then Cc would request pages from Mm, then Mm would request them back to serve its file-mapping role
Now, Mm does it all by itself. If file cahcing is requested by the FS driver, then Cc creates a file mapping and uses that to serve its purpose.

This is a rewrite of Cc
2021-02-03 09:41:22 +01:00
Jérôme Gardou 8631e75837 [NTOS:CC] Acquire the master lock after freeing the VACB in CcRosFlushDirtyPages
Fixes a random ASSERT
2020-12-09 18:06:42 +01:00
Victor Perevertkin 5c7ce4475e
[REACTOS] Cleanup INIT and some PAGE section allocations
- Change INIT_FUNCTION and INIT_SECTION to CODE_SEG("INIT") and DATA_SEG("INIT") respectively
- Remove INIT_FUNCTION from function prototypes
- Remove alloc_text pragma calls as they are not needed anymore
2020-11-02 21:45:31 +03:00
Serge Gautherie 81be201d98
[NTOS:CC] CcMdl*Complete(): Return if FastIO succeeded (#3303)
CORE-17342
2020-10-31 15:18:02 +01:00
Jérôme Gardou 1c528cbf84 Revert "[NTOS/MM]
- Fix PFNs tracing
     - Add private pages to the process working set"

This reverts commit 4c5351bf55.
Not ready for prime time
2020-10-20 15:56:21 +02:00
Jérôme Gardou 4c5351bf55 [NTOS/MM]
- Fix PFNs tracing
 - Add private pages to the process working set
2020-10-20 15:20:59 +02:00
Serge Gautherie b20f815126
[NTOSKRNL] Place INIT_FUNCTION before the return type (#2823)
(but after 'static' or SAL 2 annotation.)
Follow-up to 71fefa32, which mentions that it's actually required by the compiler in some circumstances.
2020-05-23 15:56:10 +02:00
Serge Gautherie a0d2170729 [NTOS:CC] CcWriteVirtualAddress(): Do not round 'Size'
Revert 'Properly align VACB writes' part of 2a80ae2bb6.

CORE-15384 CORE-16018
2019-07-01 21:30:25 +02:00
Pierre Schweitzer 77b6899d89
[NTOSKRNL] Don't set VACB dirty on release if already dirty
CORE-15954
2019-04-20 11:23:35 +02:00
Pierre Schweitzer bd75947372
[NTOSKRNL] Drop the no longer needed Dirty field
CORE-15954
2019-04-18 08:14:41 +02:00
Pierre Schweitzer 8b6696fcdb
[NTOSKRNL] Don't mark VACB dirty on unpin
This could happen if BCB was marked dirty previously.
Marking VACB dirty on unpin could lead to a double write of
the VACB, even if clean.
Indeed, now that setting BCB dirty leads to marking VACB
dirty, the VACB can be flushed in between by the lazy-writer.
The BCB state is not reset on VACB flush, contrary to the VACB state.
Thus, on unpin even if the VACB was already flushed, we were
setting back the dirty state, leading the VACB to be flushed again.

This could bring a small performance downgrade. Though it remains
limited since this is mostly used for FS metadata.
Possibly it could lead to metadata corruption, but this is likely
less possible.

CORE-15954
2019-04-17 22:35:19 +02:00
Timo Kreuzer 71fefa32db
[NDK][NTOS] Add global definition of INIT_FUNCTION/INIT_SECTION (#779)
* Add an NDK header to define INIT_FUNCTION/INIT_SECTION globally
* Use _declspec(allocate(x)) and _declspec(code_seg(x)) on MSVC versions that support it
* Use INIT_FUNCTION on functions only and INIT_SECTION on data only (required by MSVC)
* Place INIT_FUNCTION before the return type (required by MSVC)
* Make sure declarations and implementations share the same modifiers (required by MSVC)
* Add a global linker option to suppress warnings about defined but unused INIT section
* Merge INIT section into .text in freeldr
2018-12-30 12:19:11 +01:00
Pierre Schweitzer e16241255d
[NTOSKRNL] Set the MemoryMaker flag when executing write behind 2018-12-23 14:59:42 +01:00
Pierre Schweitzer 7e97071c8b
[NTOSKRNL] Implement write behind in Cc
For now, this is just a split between scan and flush that
were both done during lazy scan previously.
Lazy scan shouldn't perform any write operation, but only
queue a write behind operation.

Our implementation is far from the original, as it seems
our lazy scan should queue a write behind operation per
shared cache map. Right now, we only perform global
operation.
2018-12-23 14:45:38 +01:00
Pierre Schweitzer 1a267045f8
[NTOSKRNL] Honor files that shouldn't be lazy written 2018-12-23 12:10:58 +01:00
Pierre Schweitzer 0917c64812
[NTOSKRNL] When growing a file, invalid the last VACB so that it can be refreshed
This will avoid corruption when a file size is little grown and read afterwards.
Up to now, FSD where reading 0es instead of expected data, causing corruption.

This fixes MS FastFAT not being able to mount a FAT volume in ReactOS, corrupting
the FAT.
This also fixes the CcSetFileSizes kmtest tests.

This is based on a patch by Thomas Faber.

CORE-11819
2018-12-23 11:19:14 +01:00