- Wine-gitlab - list.winehq.org

[PATCH v12 0/4] MR852: ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional pread.
by Witold Baryluk (＠baryluk) 18 Sep '22

18 Sep '22

Apex Legends game periodically (every 30 seconds) calls this function with up to 22k virtual addresses. All but 1 of them is valid. Due to amount of queries addresses, and cost of seek+read, this causes this function to take up to about 50ms. So framerate drops from ~150 FPS to 20FPS for about a second. As far as I can see, returning 0 entries from this function, still makes Apex Legend work. But keep code correct, and optimise it by: 1. Opening pagemap file once, and never closing it. This eliminates repeated fopen/fseek/fread/fclose sequences, which helps even in queries with small amount of virtual addresses. 2. Using pread, instead of seek+read. 3. Only performing pagemap read when the address is valid. Future work might recognize continues pages in the query, and perform a batch read of multiple pagemap entries, instead one page at a time, but for now it is not necassary. This change get_working_set_ex peek wall clock runtime from 57ms to 0.29ms. Tested on Linux, but similar change done for the BSD part. `Signed-off-by: Witold Baryluk <witold.baryluk(a)gmail.com>` -- v12: ntdll: Keep pagemap file open after first use of NtQueryVirtualMemory(MemoryWorkingSetExInformation) https://gitlab.winehq.org/wine/wine/-/merge_requests/852

3 6

[PATCH 0/1] MR879: oledb32/tests: Free returned property set memory
by Alistair Leslie-Hughes (＠alesliehughes) 18 Sep '22

18 Sep '22

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=36312 -- https://gitlab.winehq.org/wine/wine/-/merge_requests/879

2 1

[PATCH v11 0/4] MR852: ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional pread.
by Witold Baryluk (＠baryluk) 17 Sep '22

17 Sep '22

Apex Legends game periodically (every 30 seconds) calls this function with up to 22k virtual addresses. All but 1 of them is valid. Due to amount of queries addresses, and cost of seek+read, this causes this function to take up to about 50ms. So framerate drops from ~150 FPS to 20FPS for about a second. As far as I can see, returning 0 entries from this function, still makes Apex Legend work. But keep code correct, and optimise it by: 1. Opening pagemap file once, and never closing it. This eliminates repeated fopen/fseek/fread/fclose sequences, which helps even in queries with small amount of virtual addresses. 2. Using pread, instead of seek+read. 3. Only performing pagemap read when the address is valid. Future work might recognize continues pages in the query, and perform a batch read of multiple pagemap entries, instead one page at a time, but for now it is not necassary. This change get_working_set_ex peek wall clock runtime from 57ms to 0.29ms. Tested on Linux, but similar change done for the BSD part. `Signed-off-by: Witold Baryluk <witold.baryluk(a)gmail.com>` -- v11: ntdll: Keep pagemap file open after first use of NtQueryVirtualMemory(MemoryWorkingSetExInformation) https://gitlab.winehq.org/wine/wine/-/merge_requests/852

2 5

[PATCH v10 0/4] MR852: ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional pread.
by Witold Baryluk (＠baryluk) 17 Sep '22

17 Sep '22

Apex Legends game periodically (every 30 seconds) calls this function with up to 22k virtual addresses. All but 1 of them is valid. Due to amount of queries addresses, and cost of seek+read, this causes this function to take up to about 50ms. So framerate drops from ~150 FPS to 20FPS for about a second. As far as I can see, returning 0 entries from this function, still makes Apex Legend work. But keep code correct, and optimise it by: 1. Opening pagemap file once, and never closing it. This eliminates repeated fopen/fseek/fread/fclose sequences, which helps even in queries with small amount of virtual addresses. 2. Using pread, instead of seek+read. 3. Only performing pagemap read when the address is valid. Future work might recognize continues pages in the query, and perform a batch read of multiple pagemap entries, instead one page at a time, but for now it is not necassary. This change get_working_set_ex peek wall clock runtime from 57ms to 0.29ms. Tested on Linux, but similar change done for the BSD part. `Signed-off-by: Witold Baryluk <witold.baryluk(a)gmail.com>` -- v10: ntdll: Keep pagemap file open after first use of NtQueryVirtualMemory(MemoryWorkingSetExInformation) https://gitlab.winehq.org/wine/wine/-/merge_requests/852

2 5

[PATCH v9 0/4] MR852: ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional pread.
by Witold Baryluk (＠baryluk) 17 Sep '22

17 Sep '22

Apex Legends game periodically (every 30 seconds) calls this function with up to 22k virtual addresses. All but 1 of them is valid. Due to amount of queries addresses, and cost of seek+read, this causes this function to take up to about 50ms. So framerate drops from ~150 FPS to 20FPS for about a second. As far as I can see, returning 0 entries from this function, still makes Apex Legend work. But keep code correct, and optimise it by: 1. Opening pagemap file once, and never closing it. This eliminates repeated fopen/fseek/fread/fclose sequences, which helps even in queries with small amount of virtual addresses. 2. Using pread, instead of seek+read. 3. Only performing pagemap read when the address is valid. Future work might recognize continues pages in the query, and perform a batch read of multiple pagemap entries, instead one page at a time, but for now it is not necassary. This change get_working_set_ex peek wall clock runtime from 57ms to 0.29ms. Tested on Linux, but similar change done for the BSD part. `Signed-off-by: Witold Baryluk <witold.baryluk(a)gmail.com>` -- v9: ntdll: Keep pagemap file open after first use of NtQueryVirtualMemory(MemoryWorkingSetExInformation) ntdll: Use pread in NtQueryVirtualMemory(MemoryWorkingSetExInformation) ntdll: Do not use hardcoded page shift in NtQueryVirtualMemory(MemoryWorkingSetExInformation) ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional page check https://gitlab.winehq.org/wine/wine/-/merge_requests/852

3 6

Re: [PATCH v8 0/4] MR852: ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional pread.
by Witold Baryluk (＠baryluk) 17 Sep '22

17 Sep '22

On Sat Sep 17 17:00:54 2022 +0000, Witold Baryluk wrote: > Fixed. Good eye! Hmm. I pushed to my branch https://gitlab.winehq.org/baryluk/wine/-/commit/47f155ee2ec1da7a7f62ad1ca0c… , but this MR still shows the previous content of the branch. Weird. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/852#note_8745

1 0

Re: [PATCH v8 0/4] MR852: ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional pread.
by Witold Baryluk (＠baryluk) 17 Sep '22

17 Sep '22

On Sat Sep 17 15:38:35 2022 +0000, Jinoh Kang wrote: > ```suggestion:-0+0 > if (pagemap_fd == -1 || > ``` Fixed. Good eye! -- https://gitlab.winehq.org/wine/wine/-/merge_requests/852#note_8743

1 0

Re: [PATCH v8 0/4] MR852: ntdll: Speed up NtQueryVirtualMemory(MemoryWorkingSetExInformation) by conditional pread.
by Jinoh Kang (＠iamahuman) 17 Sep '22

17 Sep '22

Jinoh Kang (@iamahuman) commented about dlls/ntdll/unix/virtual.c: > get_committed_size( view, p->VirtualAddress, &vprot, VPROT_COMMITTED ) && > (vprot & VPROT_COMMITTED)) > { > - if (!f || fseek( f, ((UINT_PTR)p->VirtualAddress >> page_shift) * sizeof(pagemap), SEEK_SET ) == -1 || > - fread( &pagemap, sizeof(pagemap), 1, f ) != 1) > + if (pagemap_fd != -1 || ```suggestion:-0+0 if (pagemap_fd == -1 || ``` -- https://gitlab.winehq.org/wine/wine/-/merge_requests/852#note_8741

1 0

[PATCH 0/1] MR877: po: Update Lithuanian translation.
by Aurimas Fišeras 17 Sep '22

17 Sep '22

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/877

1 1

Re: [PATCH v21 0/1] MR778: ntoskrnl: KeInsertQueueDpc stub
by Etaash Mathamsetty (＠etaash.mathamsetty) 16 Sep '22

16 Sep '22

On Fri Sep 16 19:50:17 2022 +0000, Etaash Mathamsetty wrote: > yes I know that, but when I clicked the rebase button the pipeline > started failing for no reason, now I have to rebase locally which I understand Yeah I have absolutely no idea why it's failing, it works on my machine with the exact same code broken pipeline for sure -- https://gitlab.winehq.org/wine/wine/-/merge_requests/778#note_8735

1 0