January 2022 - Wine-Devel - list.winehq.org

[PATCH 1/3] wined3d: Use wined3d_buffer_load_location() in wined3d_buffer_get_memory().

by Zebediah Figura

Signed-off-by: Zebediah Figura <zfigura(a)codeweavers.com> --- dlls/wined3d/buffer.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/dlls/wined3d/buffer.c b/dlls/wined3d/buffer.c index b42c7bd919a..3d66af3e5e6 100644 --- a/dlls/wined3d/buffer.c +++ b/dlls/wined3d/buffer.c @@ -653,14 +653,12 @@ DWORD wined3d_buffer_get_memory(struct wined3d_buffer *buffer, struct wined3d_co if (locations & WINED3D_LOCATION_DISCARDED) { locations = ((buffer->flags & WINED3D_BUFFER_USE_BO) ? WINED3D_LOCATION_BUFFER : WINED3D_LOCATION_SYSMEM); - if (!wined3d_buffer_prepare_location(buffer, context, locations)) + if (!wined3d_buffer_load_location(buffer, context, locations)) { data->buffer_object = 0; data->addr = NULL; return 0; } - wined3d_buffer_validate_location(buffer, locations); - wined3d_buffer_invalidate_location(buffer, WINED3D_LOCATION_DISCARDED); } if (locations & WINED3D_LOCATION_BUFFER) { -- 2.34.1

Jan. 28, 2022

Re: [PATCH resend v4 2/4] wdscore: Implement CurrentIP.

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=106259 Your paranoid android. === build (build log) === error: patch failed: configure.ac:3347 error: patch failed: dlls/wdscore/Makefile.in:1 error: patch failed: dlls/wdscore/wdscore.spec:71 Task: Patch failed to apply === debian11 (build log) === error: patch failed: configure.ac:3347 error: patch failed: dlls/wdscore/Makefile.in:1 error: patch failed: dlls/wdscore/wdscore.spec:71 Task: Patch failed to apply === debian11 (build log) === error: patch failed: configure.ac:3347 error: patch failed: dlls/wdscore/Makefile.in:1 error: patch failed: dlls/wdscore/wdscore.spec:71 Task: Patch failed to apply

Jan. 28, 2022

Re: [PATCH 7/7] winhttp: Don't allow queueing websocket receive if another is pending.

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=106257 Your paranoid android. === w8adm (32 bit report) === winhttp: notification.c:256: Test failed: failed to send request 12007 notification.c:258: Test failed: unexpected function 3, expected 4. probably some notifications were missing notification.c:260: Test failed: failed to receive response 12019 notification.c:264: Test failed: failed unexpectedly 12019 notification.c:265: Test failed: request failed unexpectedly 0 notification.c:268: Test failed: unexpected function 3, expected 13. probably some notifications were missing notification.c:116: Test failed: 268: expected status 0x00000002 got 0x00000800 notification.c:117: Test failed: 268: expected function 3 got 13 notification: Timeout

Jan. 28, 2022

Re: [PATCH 6/7] winhttp: Don't allow socket transfers after receiving close frame.

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=106256 Your paranoid android. === w8adm (32 bit report) === winhttp: notification.c:256: Test failed: failed to send request 12007 notification.c:258: Test failed: unexpected function 3, expected 4. probably some notifications were missing notification.c:260: Test failed: failed to receive response 12019 notification.c:264: Test failed: failed unexpectedly 12019 notification.c:265: Test failed: request failed unexpectedly 0 notification.c:268: Test failed: unexpected function 3, expected 13. probably some notifications were missing notification.c:116: Test failed: 268: expected status 0x00000002 got 0x00000800 notification.c:117: Test failed: 268: expected function 3 got 13 notification: Timeout

Jan. 28, 2022

Re: [PATCH 5/7] winhttp: Don't queue socket_close() in WinHttpWebSocketClose() if close frame is already received.

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=106255 Your paranoid android. === w8adm (32 bit report) === winhttp: notification.c:256: Test failed: failed to send request 12007 notification.c:258: Test failed: unexpected function 3, expected 4. probably some notifications were missing notification.c:260: Test failed: failed to receive response 12019 notification.c:264: Test failed: failed unexpectedly 12019 notification.c:265: Test failed: request failed unexpectedly 0 notification.c:268: Test failed: unexpected function 3, expected 13. probably some notifications were missing notification.c:116: Test failed: 268: expected status 0x00000002 got 0x00000800 notification.c:117: Test failed: 268: expected function 3 got 13 notification: Timeout === debian11 (64 bit WoW report) === winhttp: notification.c:116: Test failed: 994: expected status 0x00000008 got 0x00200000 notification: Timeout

Jan. 28, 2022

Re: [PATCH 3/7] winhttp: Don't mind socket state in WinHttpWebSocketQueryCloseStatus().

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=106253 Your paranoid android. === w8adm (32 bit report) === winhttp: notification.c:256: Test failed: failed to send request 12007 notification.c:258: Test failed: unexpected function 3, expected 4. probably some notifications were missing notification.c:260: Test failed: failed to receive response 12019 notification.c:264: Test failed: failed unexpectedly 12019 notification.c:265: Test failed: request failed unexpectedly 0 notification.c:268: Test failed: unexpected function 3, expected 13. probably some notifications were missing notification.c:116: Test failed: 268: expected status 0x00000002 got 0x00000800 notification.c:117: Test failed: 268: expected function 3 got 13 notification: Timeout

Jan. 28, 2022

Re: [PATCH 2/7] winhttp/tests: Test shutting down websocket when receive is pending.

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=106252 Your paranoid android. === w8adm (32 bit report) === winhttp: notification.c:256: Test failed: failed to send request 12007 notification.c:258: Test failed: unexpected function 3, expected 4. probably some notifications were missing notification.c:260: Test failed: failed to receive response 12019 notification.c:264: Test failed: failed unexpectedly 12019 notification.c:265: Test failed: request failed unexpectedly 0 notification.c:268: Test failed: unexpected function 3, expected 13. probably some notifications were missing notification.c:116: Test failed: 268: expected status 0x00000002 got 0x00000800 notification.c:117: Test failed: 268: expected function 3 got 13 notification: Timeout

Jan. 28, 2022

Re: [PATCH v4 1/4] wdscore: Implement CurrentIP.

by Mohamad Al-Jaf

Hi, Sorry to bother you all. Just wanted to get your input on this. Because CurrentIP isn't WINAPI, shouldn't __ASM_GLOBAL_FUNC be used instead of __ASM_STDCALL_FUNC? Since stdcall cleans the stack, wouldn't the stack pointer change? Or is my understanding incorrect?

Jan. 28, 2022

Re: [PATCH v4 1/4] wdscore: Implement CurrentIP.

by Mohamad Al-Jaf

This is supposed to be v5, not v4. I always check the patches on my machine and send the patches to my email to check again. This time I waited an hour before submitting and checked it again before submitting. And yet I still made a mistake, it's very frustrating. I guess I got confused with the previous v3 revision of wdscore: Add stubs for WdsSetupLogMessage[AW] and mistakenly updated all of them to v4. Terribly sorry about that. I don't know why I keep making mistakes like these.

Jan. 27, 2022

Re: [RFC PATCH v2 2/5] server: Add a new server request "notify_async_direct_result."

by Zebediah Figura (she/her)

On 1/27/22 13:22, Jinoh Kang wrote: > On 1/28/22 04:13, Jinoh Kang wrote: >> On 1/28/22 04:05, Jinoh Kang wrote: >>> + >>> +/* Notify direct completion of async and close the wait handle */ >>> +DECL_HANDLER(notify_async_direct_result) >>> +{ >>> + struct async *async = (struct async *)get_handle_obj( current->process, req->handle, 0, &async_ops ); >>> + >>> + if (!async) return; >>> + >>> + if (async->iosb && async->unknown_status && !async->pending && async->terminated) >>> + { >>> + /* Reactivate async. We call async_reselect() later. */ >>> + async->terminated = 0; >>> + >>> + /* Set result for async_handoff(). */ >>> + set_error( req->status ); >>> + async->iosb->result = req->information; >>> + >>> + /* The async_handoff() call prior to the current server request was >>> + * effectively a no-op since async->unknown_status is 1. Calling it >>> + * again with async->unknown_status = 0 will do the remaining steps. >>> + */ >>> + async->unknown_status = 0; >>> + async_handoff( async, NULL, 0 ); >>> + >>> + if (get_error() == STATUS_PENDING) >>> + { >>> + async_reselect( async ); >>> + } >>> + else if (NT_ERROR( get_error() )) >>> + { >>> + /* synchronous I/O failure: don't invoke callbacks, only dequeue it. */ >>> + async_dequeue( async ); >> >> Note: we can't use async_set_result() here. async_handoff() leaves async->terminated as 0, and async_set_result() has "assert( async->terminated )". >> >> Thus, reusing async_set_result() here requires either: >> >> 1. Setting async->pending to 0 so that async_handoff() will call async_terminate(). >> This is obviously incorrect since unwanted IOCP packets and APCs will fire on synchronous failure. >> >> 2. Not using async_handoff(). We have to copy a lot of code (particulary setting pending and direct_result flags) out of async_handoff() to do this. > > Maybe just merge both branches (error/success) and force async_terminate() before calling async_set_result()? Actually, I believe the right thing to do is just to set async->terminated = 1. In fact, I believe you should set async->terminated = 1 in the recv_socket request, otherwise async_waiting() will return 1 and the server will end up spinning and/or trying to wake the async. (Granted, that could be protected against by checking unknown_status, but...) There are a few different ways to look at this, but one way (and what I think makes the most sense to me) is that we're completing the async as if it had been alerted, much as we would from a kernel APC, except that we're also setting the initial status, which wasn't known yet. Hence we want to go through async_set_result(). Along these lines, if you also set async->alerted = 1, you get the STATUS_PENDING "restart the wait" behaviour for free. Which makes conceptual sense as well. I suppose this is why you mentioned direct_result in [1]—i.e. we could just async_terminate(STATUS_ALERTED) in recv_socket. I can see both approaches working, but I think that open-coding the relevant parts of async_terminate would be better than trying to get direct_result manipulation exactly right. Of course, I also haven't tried... [1] https://www.winehq.org/pipermail/wine-devel/2022-January/205774.html

Jan. 27, 2022