07303: coco3 [tharogad]: The Castle of Tharogad game cartridge does not run

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
07303	Media Support	Major	Always	Apr 26, 2019, 02:37	Sep 2, 2021, 01:51

Tester	star2root	View Status	Public	Platform	MAME (Self-compiled)
Assigned To	AJR	Resolution	Fixed	OS	MacOS X
Status [?]	Resolved			Driver	trs/coco3.cpp
Version	0.209	Fixed in Version	0.230	Build	64-bit
		Fixed in Git Commit	91fc427	Github Pull Request #

Summary	07303: coco3 [tharogad]: The Castle of Tharogad game cartridge does not run
Description	The rom pak for Castle of Tharogad drop you into Extended Color BASIC instead of launching the game. This pak worked in release 0.192 and is currently broken. I am pretty sure it stopped working when the support for 32k CoCo 3 game paks was added. Other 16k rom like Thexder still work, and all the 32k roms I have tried work, so this may be the only one not working. The disk/hacked version of Tharogad still runs fine.
Steps To Reproduce	Run: mame64 coco3 tharogad
Additional Information	The reason it's not work is probably related to this code update: --- src/mame/video/gime.cpp 2019-04-25 08:57:14.000000000 -0400 +++ src/mame/video/gime.cpp 2019-04-25 09:03:33.000000000 -0400 @@ -97,6 +97,7 @@ #define LOG_INT_MASKING 0 #define LOG_GIME 0 #define LOG_TIMER 0 +#define LOG_PALETTE 0 @@ -558,12 +559,15 @@ // we're in ROM static const uint8_t rom_map[4][4] = { - { 0, 1, 6, 7 }, - { 0, 1, 6, 7 }, + { 0, 1, 4, 5 }, + { 0, 1, 4, 5 }, { 0, 1, 2, 3 }, - { 4, 5, 6, 7 } + { 6, 7, 4, 5 } }; + // Pin ROM page to MMU slot + block = (block & 0xfc) \| (bank & 0x03); + // look up the block in the ROM map block = rom_map[m_gime_registers[0] & 3][(block & 0x3F) - 0x3C]; @@ -1011,6 +1015,10 @@ { offset &= 0x0F; + // perform logging + if (LOG_PALETTE) + logerror("%s: CoCo3 Palette: $%04x <== $%02x\n", describe_context(), offset + 0xffB0, data); + /* has this entry changed? / if (m_palette_rotated[m_palette_rotated_position][offset] != data) { --- src/mame/machine/coco_vhd.cpp 2019-04-25 08:56:28.000000000 -0400 +++ src/mame/machine/coco_vhd.cpp 2019-04-25 08:56:43.000000000 -0400 @@ -48,7 +48,7 @@ CONSTANTS *************************************************************************/ -#define VERBOSE 1 +#define VERBOSE 0 #define VHDSTATUS_OK 0x00 #define VHDSTATUS_NO_VHD_ATTACHED 0x02 --- src/mame/machine/6883sam.cpp 2019-04-25 08:52:28.000000000 -0400 +++ src/mame/machine/6883sam.cpp 2019-04-25 08:54:20.000000000 -0400 @@ -151,6 +151,9 @@ / if we're configuring a bank that never changes, update it now */ switch(bank) { + case 3: + m_space_C000.point(m_banks[3], m_banks[3].m_memory_offset); + break; case 4: m_space_FF00.point(m_banks[4], 0x0000); break; --- src/devices/bus/coco/coco_pak.cpp 2019-04-25 08:51:06.000000000 -0400 +++ src/devices/bus/coco/coco_pak.cpp 2019-04-25 08:51:45.000000000 -0400 @@ -199,7 +199,7 @@ uint32_t coco_pak_banked_device::get_cart_size() { - return 0x4000; + return 0x20000; } //-------------------------------------------------
Github Commit

Flags
Regression Version	0.203
Affected Sets / Systems	coco3 [tharogad]

Attached Files

No.16389 star2root Viewer Apr 28, 2019, 01:57	Removing these lines from src/mame/video/gime.cpp fixes this issue and as far as I can tell does not break any other cartridges that currently work. - // Pin ROM page to MMU slot - block = (block & 0xfc) \| (bank & 0x03); The rampage cartridge doesn't seem to work, but as far as I can tell, it never has worked and doesn't work with or without this change. All the other cartridges that I have tested seem to work as well as they did before, and this change allows tharogad to work.
No.16390 star2root Viewer Apr 28, 2019, 02:12	Looks like Super Pitfall is also broken, but also seems to be a different problem, and I see it has been reported. So removing the two lines above still doesn't appear to break anything that is currently working, and seems to fix the Castle of Tharogad issue.
No.16418 star2root Viewer May 4, 2019, 09:08	So, any comments on removing those two lines of code? I think that is the correct fix but I could be wrong.
No.16419 Tafoid Administrator May 4, 2019, 11:12	Best thing to do it push the code via a PR to the MAME repository for examination and comment. https://github.com/mamedev/mame
No.16424 star2root Viewer May 4, 2019, 21:02	Is there a way to create a pull request without creating and publicly posting a fork of the entire code base? Because I don't want to do that, but I will happily share patches.
No.18545 star2root Viewer Mar 14, 2021, 20:03	https://github.com/mamedev/mame/pull/7862/commits/66769187b1b6cadd77e825aae1253c007191ca55
No.18553 AJR Developer Mar 15, 2021, 14:45	The cause of this bug is the initialization routine executed when the program first enters the cartridge ROM: C000: ORCC #$50 C002: CLR $FFD9 C005: LDD #$3E3F C008: STD $FFA4 C00B: JMP $800E Execution then needs to continue with code from the same cartridge area having been banked where it jumps to.
No.19201 star2root Viewer Sep 2, 2021, 00:10	Have any fixes for this issue been applied or is this still broken in the current version in the main branch?
No.19204 Tafoid Administrator Sep 2, 2021, 01:51	Works fine here in 0.235 and current GIT. It was resolved back in 0.230, as the bug report states and links to the fixing commit.