Tuesday, 2022-10-25

*** openpowerbot <openpowerbot!~openpower@94-226-188-34.access.telenet.be> has quit IRC00:46
*** openpowerbot <openpowerbot!~openpower@94-226-188-34.access.telenet.be> has joined #libre-soc00:55
programmerjakecesar: you have mail delivery turned off for libre-soc-isa, that's likely caused by libre-soc.org not sending to gmail correctly over the last week (now fixed, thanks alain!):01:11
programmerjakeyou'll want to re-enable mail delivery if you didn't intentionally disable it01:14
*** openpowerbot <openpowerbot!~openpower@94-226-188-34.access.telenet.be> has quit IRC01:28
*** openpowerbot <openpowerbot!~openpower@94-226-188-34.access.telenet.be> has joined #libre-soc01:45
*** openpowerbot <openpowerbot!~openpower@94-226-188-34.access.telenet.be> has joined #libre-soc01:45
*** jab <jab!~jab@user/jab> has quit IRC02:55
*** jn <jn!~quassel@user/jn/x-3390946> has quit IRC03:43
*** jn <jn!~quassel@2a02:908:1066:b7c0:20d:b9ff:fe49:15fc> has joined #libre-soc03:43
*** jn <jn!~quassel@user/jn/x-3390946> has joined #libre-soc03:43
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has quit IRC05:25
*** openpowerbot <openpowerbot!~openpower@94-226-188-34.access.telenet.be> has quit IRC05:25
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has joined #libre-soc05:39
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC07:12
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:13
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has joined #libre-soc08:19
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC08:36
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc08:37
*** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC09:31
*** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc10:35
lkclmarkos, you removed the payments from #229. https://bugs.libre-soc.org/show_activity.cgi?id=22910:59
markoswhen did I do that11:00
lkclkonstantinos@vectorcamp.gr2022-10-14 11:44:09 BST11:00
markosI think I might have added a comment and didn't refresh the page when I did11:00
*** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC11:01
markosno i didn't add anything11:01
markosI have no idea how this happened11:01
lkcland you got a warning, "your changes may overwrite other edits"?11:01
lkcland you went, "yeah i can ignore that"?11:01
markosyeah I added a comment11:02
lkclnow you know, you can't ignore the warnings :)11:02
markosI probably clicked so fast I missed the warning11:02
lkclparticularly as that's for financial Auditing (EU Compliance for NLnet)11:03
markosI'll be more careful next time :D11:03
lkclfortunately the history contains what was in it so i've restored it11:03
markosthanks and again sorry11:03
lkcl(and also updated the payment date information for you)11:03
lkclcan you check that you've received the money? (14th and 20th)11:03
markosyes, all good11:04
markossame day deposits even11:04
lkclyehyeh. zero charges to NLnet as it's EU bank transfers11:05
lkclahh now i have the fun task of reconciling 5 remaining grants that ended mid-october11:06
lkclremoving all outstanding payments urrrr11:06
lkclghostmansd[m], "Is Rc used to indicate the two modes?"11:19
lkclRc=1 is a post-analysis phase11:19
lkclthere's no "&" operator in Power ISA notation (there is, it's a logical-and, not bitwise-and). get the bits directly11:21
lkcl    n <- (RB)[XLEN/2:XLEN-1] # Limit RB to upper word (32-bits)11:21
lkclalso bear in mind, the Power ISA spec - and the whole reason why SelectableInt exists - is extreme-strict about bitlength11:22
lkcl(unlike programming languages which don't give a shit, with the possible exception of ADA)11:23
programmerjakeand rust...it intentionally has no silent conversions between integer types, you need to explicitly convert.12:05
programmerjakeactually, & in pseudocode notation is bitwise and: https://git.libre-soc.org/?p=openpower-isa.git;a=blob;f=openpower/isa/fixedlogical.mdwn;h=28311ce33583aec9a3eba5c5bdf20304b1fea263;hb=753ba7273a910ba697173d58ea998f7730e6849d#l2112:06
programmerjakepowerisa just calls that logical and, it's still bitwise though, not c && or python `and`12:07
programmerjakehave a safe family trip, luke!12:11
*** octavius <octavius!~octavius@> has joined #libre-soc12:31
lkclprogrammerjake, appreciated. got a "new" car, 1988 peugeot 205 1.8 diesel, 60mpg (!) for it12:40
octaviusOh wow!12:41
lkcl(the other one, the brake pads are shot, clutch is going: brake replacement == GBP 300, clutch replacement == GBP 350+, which is 50% the value of the 35-year-old peugeot 205)12:41
octaviusYou going away for the day?12:41
octaviusvery nice12:42
lkclliverpool museum.  doctor who exhibition12:42
lkcllast week, but outside of school half-term. they extended it to nov 6th12:42
octaviusAre you going straight there, or stopping somewhere in the middle?12:43
lkclstopping at least twice12:43
lkclin a 35-year-old vehicle i really don't want to go over 65 mph, which means it'll be.... at least 4 hours12:44
lkcli'm taking the laptop (etc.) and probably the UPS :)  and the WIFI 4G router...12:46
lkcli just won't be online all the time12:46
octaviusSure, but do try to spend time with the family :)12:48
octaviusAlso, I'll soon be getting one of those smarty sim's, and testing it out with a USB modem and openwrt router12:49
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has quit IRC12:51
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has joined #libre-soc13:01
lkcllet me put you in touch with marie, we get 20 quid for an introduction13:05
octaviusHopefully your link will work (as I've registered already)13:30
lkcloctavius, oh doh13:32
lkcloctavius, can you avoid inserting tabs into pages (ls003 pseudocode has been converted to tabs)13:37
octaviusSorry, not sure how that happened, I'll check better next commit13:37
lkclif you're using vim you can set up a ~/.vimrc13:39
lkclsyntax on13:40
lkclset ts=413:40
lkclset sw=413:40
lkclset expandtab13:40
lkclexpandtab's the "important" one13:40
octaviusThanks, will do13:44
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has quit IRC14:05
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has quit IRC14:05
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has joined #libre-soc14:20
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has joined #libre-soc14:20
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has quit IRC15:07
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has quit IRC15:08
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has joined #libre-soc15:21
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has quit IRC15:42
*** octavius <octavius!~octavius@> has quit IRC15:53
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has joined #libre-soc15:57
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has joined #libre-soc15:58
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC16:05
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc16:06
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC16:51
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc16:52
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC16:56
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc16:57
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC17:15
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc17:15
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC17:37
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc17:39
*** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc17:51
*** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC18:25
*** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc18:25
lkclghostmansd, looks great.18:27
lkclit is possible to do this:18:27
ghostmansdnot yet, there's some fuckup with sm operand18:27
lkcl case (0):  sum[0:XLEN-1] <- (((RB) || 0b0) + (RA))18:28
ghostmansdits range is being considered to be 318:28
lkclyou have +    switch (sh)18:28
lkclnot "switch (sm)"18:28
lkcldefinitely should not be the case18:28
lkcldsld uses a switch (sm) and it works perfectly fine18:29
ghostmansdswitch (sh)18:29
ghostmansdAh yes, 1 sec18:29
lkclseveral unit tests confirm that a switch(sm) works perfectly18:29
ghostmansdno the issue is not in the switch18:29
lkclalso you want the same as what i did as a throwaway comment yesterday18:29
ghostmansdhang on18:29
lkcln | 0b0 will be 65 bits18:30
lkcl(RB) || 0b0 will be 65 bits sorry18:30
lkcl(B) || 0b00 will be 66 bits18:30
lkcland like i said this morning, and as you know from using SelectableInt(), a 66-bit thing {any-operator-with} a 64-bit thing is guaranteed to be an error18:31
ghostmansdOK fixed sm18:31
ghostmansdnext, to bit count18:31
ghostmansddo you want [0:XLEN-1] on the other side?18:32
lkclcase (0): RT <- ((RB)[0:XLEN-2] || [0]) + (RA)18:32
ghostmansdsum[0:XLEN-1] <- (((RB) || 0b0) + (RA))[0:XLEN-1]18:32
lkclcase (1): RT <- ((RB)[0:XLEN-3] || [0]*2) + (RA)18:32
ghostmansdAh, so we discard the last bit18:32
ghostmansdRight, that I forgot18:32
lkclyes. it's essential.18:33
lkclpain-in-the-ass, but essential.18:33
lkclit's so that there's no possibility of ambiguity in the spec18:33
ghostmansdsum[0:XLEN-1] <- (((RB)[XLEN-4-1] || 0b0000) + (RA))18:33
ghostmansdsame for n18:33
lkclsum[0:XLEN-1] <- (((RB)[XLEN-4:XLEN-1] || 0b0000) + (RA))18:33
lkcl(RB)[XLEN-4-1] is of length 1 bit18:34
ghostmansdAh yes18:34
lkclfor shadduw you have to set up n <- [0]*(XLEN/2) || (RB)[XLEN/2:XLEN-1]18:34
lkclto make it 64-bit18:35
lkcl*then* you can do the add-with-shift part18:35
ghostmansdOK should be better now18:37
ghostmansdah wait18:37
ghostmansdno, from 018:37
*** octavius <octavius!~octavius@> has joined #libre-soc18:38
*** openpowerbot_ <openpowerbot_!~openpower@94-226-188-34.access.telenet.be> has quit IRC18:39
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has quit IRC18:39
ghostmansdsum[0:XLEN-1] <- (((RB)[XLEN-4:XLEN-1] || 0b0000) + (RA))18:42
ghostmansdShouldn't this also start with 0 for RB?18:42
ghostmansdplease check the recent version18:42
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has joined #libre-soc18:52
ghostmansdOK I found the issue what's wrong when building the basic opcode for shadd/shadduw19:06
ghostmansdsm overlaps with XO19:06
ghostmansdbit 2119:06
ghostmansdYeah solved this19:11
ghostmansdfor the record, this happens in the current code I have, perhaps not present on master19:12
ghostmansdanyway, these changes are not related19:12
ghostmansdwill be pushed separately19:12
lkclyes, that's deliberate.  it should be overlapping with 21 and 22.19:44
lkclremember "--11101110-,ALU,OP_SHADD,RA"?19:44
lkclthe discussion yesterday?19:44
lkclghostmansd, https://libre-soc.org/irclog/%23libre-soc.2022-10-24.log.html#t2022-10-24T21:01:5219:45
lkclrebased, all good19:47
lkclthere's nothng "wrong"... oh except it might "appear" to be "wrong" in the Z23 function, you need the bottom 2 bits of the XO to be zero19:49
lkcl    _insn("shadd",  PO=22, XO=0b01101110, Rc=0),19:49
lkclthose are effectively completely different XOs19:49
lkclthe "XO-from-the-csv-file" where the bit-pattern-start-and-end are defined in insndb.csv19:50
lkcl(as 21-31)19:50
lkclthe *manual* re-creation of Z23() it specifies bits 23-30, rather than picking up XO from the CSV file (as it will do in the future)19:51
lkclbut that's for bug... ermermermerm... 950 to solve19:52
lkclthe overlap - the fact that csv-files can have an XO which is *not* strictly part of the spec - is something that'll have to be dealt with somehow19:53
lkclthere's nothing "wrong" with what you've done, it's perfect.19:53
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has quit IRC19:59
lkclnext thing, some unit tests.  those go under... https://bugs.libre-soc.org/show_bug.cgi?id=96720:01
ghostmansd[m]Will do these soon20:02
ghostmansd[m]This _custom_insn stuff will be dropped soon, added to keep in sync with master20:03
lkclack. just putting links into bugreport20:03
ghostmansd[m]But perhaps my local branch will work even without these20:03
ghostmansd[m]Didn't have time to check it now, since that branch must be rebased20:03
ghostmansd[m]And the tasks which depend on this branch are put on hold anyway20:04
octaviuslkcl, I saw you changed "[0]*64" to "EXTZ()" in divmod2du (fair enough, I forgot that's what's used in the PowerISA spec),20:08
octaviushowever should "[0]*64" in the "if" statement be changed to "0"?20:08
lkclyes i put it back to what it was20:08
octaviusAh, is the [0]*64 syntax acceptable?20:09
lkclyou mean "if ((RA) <u (RB)) & ((RB) != [0]*64) then" ?20:09
lkclwell, have a look at the original20:09
lkclwhich has unit tests.20:09
lkclwhich work 100%.20:10
octaviusI just thought [x]*y was a Python syntax20:11
octaviusDidn't see it in the PowerISA spec20:11
lkclit's a subscript notation in Power ISA20:11
lkclwhich we can't obviously do in ASCII20:11
octaviusAh....makes sense!20:11
*** toshywoshy <toshywoshy!~toshywosh@ptr-377wf33o3bnthuddmycb.18120a2.ip6.access.telenet.be> has joined #libre-soc20:14
ghostmansdlkcl, before unit tests, I'll provide binutils support20:36
ghostmansdI'll do it in scope of 966, it'd be annoying to have the fourth task20:38
ghostmansdbinutils support turned out to be easier than I expected21:06
octaviusAh, || is concatinate... that makes a lot more sense :D21:28
*** octavius_ <octavius_!~octavius@> has joined #libre-soc21:31
*** octavius <octavius!~octavius@> has quit IRC21:34
*** jab <jab!~jab@user/jab> has joined #libre-soc21:44

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!