05153: genesis, megadriv, pico: Crash when loading a cart

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
05153	Gameplay	Critical (emulator)	Always	Feb 12, 2013, 08:22	Mar 9, 2013, 20:58

Tester	Robbbert	View Status	Public	Platform	MESS (Self-compiled)
Assigned To	wilbert	Resolution	Fixed	OS	Windows Vista/7 (64-bit)
Status [?]	Resolved			Driver	sega/mdconsole.cpp
Version	0.148u1	Fixed in Version	0.148u2	Build	Normal
		Fixed in Git Commit		Github Pull Request #

Summary	05153: genesis, megadriv, pico: Crash when loading a cart
Description	Error is : Attempted to install delegate device_md_cart_interface::read without a bound object. Happens with and without software list. Systems are currently unusable
Steps To Reproduce
Additional Information	Genesis uses a MACHINE_START, and installs 3 readwrite handlers, and a write handler. All 3 of the read handlers crash during install. Seems device_md_cart_interface::read doesn't actually exist.
Github Commit

Flags
Regression Version
Affected Sets / Systems	genesis, megadriv, pico

Attached Files

No.09340 Haze Senior Tester Feb 12, 2013, 08:41	I'm only seeing crashes on some games, others work just fine.
No.09341 Robbbert Senior Tester Feb 12, 2013, 08:50	Can you suggest one that works? With or without software list?
No.09342 Haze Senior Tester Feb 12, 2013, 09:48	99% of them, actually the only one crashing here is the one eta claimed to have fixed ( Ya Se Chuan Shuo (Chi) )
No.09344 Tafoid Administrator Feb 12, 2013, 12:36	When I last went through the list, there were about 70 or so on the megadriv softlist what would crash for various reasons and usually only when using tester type command, most notably -video none and -str. Despite the backtraces I produced for 4 different types of crashes that are happening, etabeta was unable to duplicate it on his SDL development platform. This driver works fine for most softlist items at the moment and in certainly not broken. As Haze says, most everything will work fine. If you are still have problems beyond that where everything in the whole associated list of drivers isn't working and crashing instantly - you need to check your actual box for issues.
No.09346 R. Belmont Developer Feb 12, 2013, 15:58	All the "name brand" Genny games work for me: the Sonic series, Streets of Rage, Gunstar Heroes, etc, etc.
No.09349 B2K24 Senior Tester Feb 13, 2013, 00:28	The one Haze mentions crashes here as well. Everything else in the softlist I've tried boots successfully and gets in game.
No.09352 Tafoid Administrator Feb 13, 2013, 08:55 edited on: Feb 14, 2013, 12:00	Well, I'm closing this out. The widespread issues are not able to be duplicated by anyone else. While there are some megadriv titles that don't load or crash, they should be resolved soon enough. Reopened for now since we finally got another person who could duplicate.
No.09378 Firewave Senior Tester Feb 14, 2013, 19:37	yasech crash was confirmed. ==7594== Invalid read of size 2 ==7594== at 0x138CE3F: md_rom_yasech_device::read(address_space&, unsigned int, unsigned short) (md_rom.c:552) ==7594== by 0x138CE92: non-virtual thunk to md_rom_yasech_device::read(address_space&, unsigned int, unsigned short) (md_rom.c:553) ==7594== by 0x3251773: delegate_base<unsigned short, address_space&, unsigned int, unsigned short, _noparam, _noparam>::operator()(address_space&, unsigned int, unsigned short) const (delegate.h:542) ==7594== by 0x340C0C6: handler_entry_read::read16(address_space&, unsigned int, unsigned short) const (memory.c:394) ==7594== by 0x340EF89: address_space_specific<unsigned short, (endianness_t)1, true>::read_native(unsigned int, unsigned short) (memory.c:1060) ==7594== by 0x340FA1E: unsigned char address_space_specific<unsigned short, (endianness_t)1, true>::read_direct<unsigned char, true>(unsigned int, unsigned char) (memory.c:1157) ==7594== by 0x340D67F: address_space_specific<unsigned short, (endianness_t)1, true>::read_byte(unsigned int) (memory.c:1389) ==7594== by 0x5B0B7E: delegate_base<unsigned char, unsigned int, _noparam, _noparam, _noparam, _noparam>::operator()(unsigned int) const (delegate.h:540) ==7594== by 0x2A048A9: m68ki_read_8_fc(_m68ki_cpu_core, unsigned int, unsigned int) (m68kcpu.h:1073) ==7594== by 0x2A073E0: OPER_AL_8(_m68ki_cpu_core) (m68kcpu.h:1311) ==7594== by 0x2A78215: _m68ki_cpu_core::m68k_op_cmpi_8_al(_m68ki_cpu_core) (m68kops.c:11895) ==7594== by 0x29E9B5A: cpu_execute_m68k(legacy_cpu_device) (m68kcpu.c:809) ==7594== Address 0x12389040 is 0 bytes after a block of size 2,097,152 alloc'd ==7594== at 0x80D583C: malloc (vg_replace_malloc.c:270) ==7594== by 0x3FCB834: osd_malloc_array (sdlos_unix.c:101) ==7594== by 0x32E4F8B: malloc_array_file_line(unsigned long, char const*, int) (emualloc.c:171) ==7594== by 0x138034B: device_md_cart_interface::rom_alloc(running_machine&, unsigned long) (emualloc.h:184) ==7594== by 0x1381269: base_md_cart_slot_device::load_list() (md_slot.c:369) ==7594== by 0x1380F76: base_md_cart_slot_device::call_load() (md_slot.c:336) ==7594== by 0x138112B: non-virtual thunk to base_md_cart_slot_device::call_load() (md_slot.c:357) ==7594== by 0x326732E: device_image_interface::finish_load() (diimage.c:1093) ==7594== by 0x32F717C: image_postdevice_init(running_machine&) (image.c:268) ==7594== by 0x32D6A23: driver_device::device_start() (driver.c:269) ==7594== by 0x3258195: device_t::start() (device.c:449) ==7594== by 0x33B2E01: running_machine::start_all_devices() (machine.c:960) This and the other crashes should be fixed in r21060.
No.09408 Robbbert Senior Tester Feb 22, 2013, 12:50	Made some progress. Someone fixed the crash on exit last week. The loading problem is because the last letter of the filename is getting chopped off somewhere, so it tries to load a file that doesn't exist.
No.09412 B2K24 Senior Tester Feb 23, 2013, 03:11	Ya Se Chuan Shuo (Chi) loads successfully after compiling latest SVN (21356)
No.09429 Tafoid Administrator Mar 9, 2013, 20:57	This should be resolved.