Search
lxdream.org :: lxdream/src/sh4/sh4mem.c :: log
lxdream 0.9.1
released Jun 29
Download Now
filename src/sh4/sh4mem.c
changeset931:430048ea8b71
authornkeynes
dateTue Dec 23 05:48:05 2008 +0000 (15 years ago)
permissions-rw-r--r--
last changeMore refactoring and general cleanup. Most things should be working again now.
Split off cache and start real implementation, breaking save states in the process
file annotate diff log rss
931:430048ea8b7115 yearsnkeynesMore refactoring and general cleanup. Most things should be working again now. lxdream-mem
930:07e5b11419db15 yearsnkeynesRemove pointer cache and add full address-space map. Much better lxdream-mem
929:fd8cb0c82f5f15 yearsnkeynesFirst pass experiment using cached decoding. lxdream-mem
927:17b6b9e245d815 yearsnkeynesAdd return-address-modifying exception return code to mmu TLB lookups (a little bit faster)
912:c5606ea4423215 yearsnkeynesRemove FASTCALL from mem_copy_*, not really helping atm (and sometimes hurting)
911:2f6ba75b84d115 yearsnkeynesDeclare mem_copy_* functions as FASTCALL
910:661367ef919d15 yearsnkeynesAdd shortcut test for long writes to the store queue (far and away the most popular P4 write)
905:4c17ebd9ef5e15 yearsnkeynesUse regparam calling conventions for all functions called from translated code,
828:b42865f00fb515 yearsnkeynesSilence the SCIF + SDRAM warnings for now
740:dd11269ee48b15 yearsnkeynesRationalize the two SH4 run slice impls into sh4.c, and tidy up the vm exits.
736:a02d1475ccfd15 yearsnkeynesRe-indent everything consistently
598:8798c3f0bf7816 yearsnkeynesUpdate sh4_get_region_by_vma (not that anyone uses it...)
586:2a3ba82cf24316 yearsnkeynesMerged lxdream-mmu r570:596 to trunk
577:a181aeacd6e816 yearsnkeynesRemove asm file and convert to inline (easier to cope with platform conventions) lxdream-mmu
571:9bc09948d0f216 yearsnkeynesMore MMU work in progess. Much better now... lxdream-mmu
570:d2893980fbf516 yearsnkeynesChange to generate different code for mmu on/off cases lxdream-mmu
569:a1c49e1e877616 yearsnkeynesBring icache partially into line with the mmu, a little less slow with AT off lxdream-mmu
562:e598411b640b16 yearsnkeynesAdd configure and command-line parameters to perform IO tracing lxdream-mmu
561:533f6b47807116 yearsnkeynesEnable Id keyword on all source files lxdream-mmu
559:06714bc6427116 yearsnkeynesCommit first pass at full TLB support - still needs a lot more work lxdream-mmu
550:a27e3134014716 yearsnkeynesAdd support for the MMIO side of the TLB (and LDTLB)
527:14c9489f647e16 yearsnkeynesx86-64 translator work-in-progress
504:61afb3921c4a16 yearsnkeynesFix ptr->int conversions for 64bit lxdream_0_8_1
502:c4ecae2b1b5e16 yearsnkeynesAdd sh4ptr_t type, start converting bare pointer refs to it
448:3e5f0c52f0ec16 yearsnkeynesDowngrade the errors to warnings - they're more whinges than anything fatal
430:467519b050f416 yearsnkeynesFix compilation warnings
422:61a0598e07ff16 yearsnkeynesFix compilation warnings
419:09f85e586ff216 yearsnkeynesReinstate xlat invalidation calls
418:b9b14afa095916 yearsnkeynesAdd explicit branch cases for main ram - yes it's faster...
406:9289b62f8f3316 yearsnkeynesRemove MMU check (probably shouldn't be here anyway), and disable
400:049d72a7a22916 yearsnkeynesAdd translation cache invalidation
371:f2fe152cfc9b16 yearsnkeynesFix warning
369:4b4223e7d72016 yearsnkeynesMove the store queue operation to a function in sh4mem.c
355:18bd008f700a17 yearsnkeynesBug 27: FBO support MILESTONE2
325:5717ae5d474617 yearsnkeynesMove PVR2 dma handling (0x10000000-0x13FFFFFF) into pvr2mem.c, minor
316:5ae06185a3cd17 yearsnkeynesRefactor render buffer mgmt to pvr2mem.c
282:01e53698ff3817 yearsnkeynesFirst cut of YUV converter
234:8759d0067e9d17 yearsnkeynesDisable watchpoints by default - save some cpu time and we're not really using them anyway
217:d0b3da72045617 yearsnkeynesFlush render buffer back to vram on read as well as write
172:59cf18b4cfb217 yearsnkeynesAdd register-by-register NOTRACE option (useful for tracing eg 0xA05F6000
165:126794f2ea6d17 yearsnkeynesInvalidate render buffers on writes to vram
156:3b93648a3b0717 yearsnkeynesAdd P4 I/O tracing
150:c728a442b43f17 yearsnkeynesAdd texcache invalidates on direct writes to 64-bit vram.
137:41907543d89018 yearsnkeynesCommit semi-correct G2-bus event register handling
130:279acbacc29818 yearsnkeynesTreat the 0x11000000 region as write-only texture ram (appears to be correct)
103:9b9cfc5855e018 yearsnkeynesMore rendering work in progress. Almost there now...
100:995e42e96cc918 yearsnkeynesSplit pvr2.c out to separate files for TA and renderer, minor renames
90:88e4872c2f5818 yearsnkeynesFix copy_to_sh4 for ta + vram-64 regions
83:72c4fbd6096518 yearsnkeynesFix P4 to include E000000
70:a5a5cef16deb18 yearsnkeynesInclude sh4core.h (needed to compile)
38:9ccc7ac66a9d18 yearsnkeynesRemove log lines for store queue + writes to op address cache
35:21a4be09830418 yearsnkeynesRemove modules.h - move definitions into dream.h
10:c898b37506e018 yearsnkeynesAdd CPU disasembly options to mode dropdown
.