Sunday, 2023-05-28

*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC02:27
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc02:37
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc02:54
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC06:45
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc06:46
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC08:56
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc08:59
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC09:09
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc09:12
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC11:55
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc11:56
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC13:28
*** josuah <josuah!~josuah@> has quit IRC17:04
*** josuah <josuah!~josuah@> has joined #libre-soc17:04
*** ghostmansd <ghostmansd!> has joined #libre-soc17:12
ghostmansdlkcl, your recent commits to power_enum and power_insn break binutils :-(17:14
ghostmansdpower_enums is fixed already17:14
ghostmansdas for power_insn, here's what I get upon $(python3 src/openpower/sv/ include/opcode/ppc-svp64-gen.h)17:15
ghostmansd    extra_idx_in1 = property(lambda self: self.svp64.extra_idx_in1)17:15
ghostmansdTypeError: extra_idx() missing 1 required positional argument: 'regtype17:15
ghostmansdthe thing is that there are many properties which rely on extra_idx function17:16
ghostmansdthey all just pass key ("in1", "in2", etc.)17:16
ghostmansdbut you introduced another argument, regtype17:16
ghostmansdI don't know how to organize it for sv_binutils script17:18
ghostmansdfrankly I don't even get why we now have .SRC and .DST check; I assumed *in always mean SRC and *out always mean DST17:19
ghostmansdOK looks like it's sorted now; could you please recall how to reproduce the original error?17:24
ghostmansdI think this is the one: $(echo "sv.ldu/ff=~RC1/vli *16, 0(*17)" | SILENCELOG=true pysvp64asm | powerpc64le-linux-gnu-as - -o /tmp/elf.o && powerpc64le-linux-gnu-objcopy -Obinary /tmp/elf.o /tmp/bin.o && pysvp64dis -v < /tmp/bin.o)17:37
ghostmansdYes this is it, it should have 7 bits, that's what you fixed.17:46
ghostmansdPushed both fixes into master17:46
ghostmansdI think we should at least casually check that works :-)17:48
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc18:18
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC18:22
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc18:23
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC18:25
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc18:25
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC18:30
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc18:31
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC19:10
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc19:11
ghostmansdlkcl, could you, please, explain, point me to the next task number?19:14
ghostmansdI assume that's about 2PM, but I need some starting point19:14
lkclsorry - sunday, i did hoovering and book-moving. SHP1 needs doing19:21
lkclthe original idea was that src *would* be possible to be different from dest (for the same reg).19:22
lkclthat's now gone19:22
lkclthank you for sorting those things out19:22
lkcli see the magic IRC channel is still working even if nobody is here19:22
lkclquite impressive19:22
*** yambo <yambo!> has quit IRC19:37
*** yambo <yambo!> has joined #libre-soc19:37
*** octavius <octavius!> has joined #libre-soc19:53
*** yambo <yambo!> has quit IRC20:09
ghostmansdSHP1 is almost ready20:15
ghostmansdyeah magic channel20:16
ghostmansdkinda as they have mana channels in game, lol20:16
*** yambo <yambo!> has joined #libre-soc20:22
*** octavius <octavius!> has quit IRC20:48
ghostmansdsomething is broken with gdb after the today's rebase, need time to investigate20:49
ghostmansdmedia assembly is broken, predicate macros cannot be expanded anymore20:50
*** octavius <octavius!> has joined #libre-soc21:02
ghostmansdThey committed a new routine for resolving the register names, and this one must be used.21:18
ghostmansdYes, this fixes the media tests. So much time lost in debugging just for one line of code.21:24
ghostmansdI've squashed this update into the commit which added the expansion for predicates so that it's transparent.21:25
ghostmansdlkcl, beware, we only have 1 bit in svp64_desc left.21:27
ghostmansdThat's how it looks now after regeneration:
ghostmansdhere are the changes since last generation took place:
ghostmansdptype and in3 got expanded21:29
ghostmansdI remember you wanted to keep this no larger than 64-bit integer, so we're walking over the thin ice21:30
ghostmansdIs something changed again with /pi??21:37
ghostmansdimplicit declaration of function 'svp64_insn_get_prefix_rm_ldst_imm_post_pi'; did you mean 'svp64_insn_get_prefix_rm_ldst_imm_simple_pi'?21:37
ghostmansdThis is so annoying, openpower-isa changes way too often21:38
ghostmansdYes I see, pi and lf were moved into LDST Idx/Imm simple modes21:39
ghostmansdYou know these also have to move to binutils, right?...21:40
ghostmansdI cannot really even check whether the generated code works, because each and every generation leads to massive set of changes...21:41
ghostmansdIs it at least close to final revision?21:41
ghostmansdAll I wanted is to change that shadd is changed to sadd, damn :-)21:41
ghostmansdOTOH, I like how LDST Idx and LDST Imm now look. They now follow almost the same logic, except for simple mode.21:49
ghostmansdAh damn, tables are changed too...22:02
ghostmansdsat mode is lost, therefore the updates for tables too22:02
ghostmansdSigh, now ff3/ff522:05
ghostmansdclass CROpFF5RM: RC1 = 122:08
ghostmansd[m]Only just realized that my VM (which was suspended) still thinks it's May 15. Don't be shocked by commit dates. :-)22:12
ghostmansd[m]And, if possible, fix these, please.22:12

Generated by 2.17.1 by Marius Gedminas - find it at!