07888: warrior: [debug] Assert after -bench or -str run

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
07888	Crash/Freeze	Critical (emulator)	Always	Mar 1, 2021, 01:19	1 day ago

Tester	Tafoid	View Status	Public	Platform	MAME (Self-compiled)
Assigned To		Resolution	Open	OS	Windows Vista/7/8 (64-bit)
Status [?]	Confirmed			Driver	cinematronics/cinemat.cpp
Version	0.229	Fixed in Version		Build	Debug
		Fixed in Git Commit		Github Pull Request #

Summary	07888: warrior: [debug] Assert after -bench or -str run
Description	Assertion failed: PRIMFLAG_GET_BLENDMODE(prim.flags) == BLENDMODE_NONE \|\| PRIMFLAG_GET_BLENDMODE(prim.flags) == BLENDMODE_ALPHA, file ../../../../../src/emu/rendersw.hxx, line 528
Steps To Reproduce
Additional Information
Github Commit

Flags	Debug build specific
Regression Version	0.223
Affected Sets / Systems	warrior

Attached Files	warrior_lay_works.zip (554 bytes) 2 days ago Uploaded by hap
Attached Files	default_test.zip (427 bytes) 1 day ago Uploaded by hap

No.20102 Robbbert Senior Tester Apr 28, 2022, 10:36	Tested just now with debug build, no crash
No.20119 Tafoid Administrator Apr 28, 2022, 20:27 edited on: Apr 28, 2022, 20:28	Still happens for me with 0.242 Debug build and release243 branch Debug build. The line number changed but the issue is the same. [MINGW64] R:\mame243>mamed warrior -bench 10 stream Default Resampler ':mono' changing rates 48000 -> 11025 stream Default Resampler ':soundboard' changing rates 48000 -> 11025 Speaker ':mono' Ch.0 @ 11025 Default Resampler ':mono' Ch.0 @ 11025 Warrior Sound Board ':soundboard' Ch.0 @ 11025 Default Resampler ':soundboard' Ch.0 @ 11025 Netlist Sound Device ':soundboard:sound_nl' Ch.0 @ 48000 Assertion failed: PRIMFLAG_GET_BLENDMODE(prim.flags) == BLENDMODE_NONE \|\| PRIMFLAG_GET_BLENDMODE(prim.flags) == BLENDMODE_ALPHA, file R:/mame243/src/emu/rendersw.hxx, line 596 This application has requested the Runtime to terminate it in an unusual way. Please contact the application's support team for more information.
No.20122 Robbbert Senior Tester Apr 29, 2022, 05:07	Yes, I can reproduce it now mamed warrior -bench 10 -artpath artwork will crash as you say. But if there's some external artwork, it seems much happier.
No.20124 Robbbert Senior Tester Apr 29, 2022, 05:58	The problem appears to lie in the internal artwork. It was created by Aaron (which is where this bug started), and then updated by Vas 6 weeks later (and the bug was not detected). Removing the artwork from cinemat.cpp gets rid of the assert, but I'm not an artwork expert - I cannot see anything wrong with it.
No.22453 hap Developer 2 days ago	On a Windows normal build (aka the MAME release .exe), you can repro it like: mame.exe warrior -video gdi result = no internal artwork It appears it doesn't like this: <screen index="0" blend="add"> But a lot of internal artwork uses this blendmode on stuff, and it works fine? If you draw the internal artwork the other way around, it'll also work fine, I'll attach a file. If noone wants to fix the underlying issue, that can be used.
No.22455 hap Developer 1 day ago	I attached another test file. put default.lay in artwork/pacman put default.lay in artwork/tempest on MS Windows: mame pacman -video gdi mame tempest -video gdi pacman works, tempest doesn't show the internal artwork
No.22458 cuavas Administrator 1 day ago	So is the actual issue that the software compositor doesn't support support add and multiply blend modes for vector primitives?
No.22460 hap Developer 1 day ago	At the time it reaches the assert during this test, PRIMFLAG_GET_BLENDMODE(prim.flags) == BLENDMODE_ADD. Yes, blend="multiply" also won't work there. FWIW the test case with pacman also fails with multiply blending, draw_quad_palette16_multiply is unimplemented, it gives a fatalerror (draw_quad_yuy16_multiply is missing too)