Hi. I've been tinkering again with the dist I shared before in http://adr.freeshell.org/plan9, basically 9legacy + Miller's 9pi + 9front's libseq. Importing ori's git9 I noticed that the compiler was truncating a constant without the correspondent suffix. According to C99, the type of a constant should be the first in which its value can be represented following this table: Suffix Decimal 0 or 0x ---------------------------------------------------------- none int int long int unsigned int long long int long int unsigned long int unsigned long long int ---------------------------------------------------------- u|U unsigned int unsigned int unsigned long int unsigned long int unsigned long long int unsigned long long int ---------------------------------------------------------- l|L long int long int long long int unsigned long int long long int unsigned long long int ---------------------------------------------------------- u|U & l|L unsigned long int unsigned long int unsigned long long int unsigned long long int ---------------------------------------------------------- ll|LL long long int long long int unsigned long long int ---------------------------------------------------------- u|U & ll|LL unsigned long long int ---------------------------------------------------------- Which follows the K&R description at "A.2.5.1 Integer Constants", just adding LL. Now, in plan9 constants are of type int if there is no suffix, or of the one specified by the suffix. The only change made is from a signed type to an unsigned one if it is necessary to fit the constant's value. The rest is truncated: /sys/src/cmd/cc/lex.c: [...] vv = yylval.vval; if(c1 & Numvlong) { if((c1 & Numuns) || convvtox(vv, TVLONG) < 0) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong) { if((c1 & Numuns) || convvtox(vv, TLONG) < 0) { c = LULCONST; t = TULONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } if((c1 & Numuns) || convvtox(vv, TINT) < 0) { c = LUCONST; t = TUINT; goto nret; } c = LCONST; t = TINT; goto nret; nret: yylval.vval = convvtox(vv, t); if(yylval.vval != vv){ nearln = lineno; warn(Z, "truncated constant: %T %s", types[t], symb); } return c; [...] 9front introduces some widening: [...] vv = yylval.vval; /* * c99 is silly: decimal constants stay signed, * hex and octal go unsigned before widening. */ w = 32; if((c1 & (Numdec|Numuns)) == Numdec) w = 31; if(c1 & Numvlong || (c1 & Numlong) == 0 && (uvlong)vv >= 1ULL<<w){ if((c1&(Numdec|Numvlong)) == Numdec && vv < 1ULL<<32) warn(Z, "int constant widened to vlong: %s", symb); if((c1 & Numuns) || convvtox(vv, TVLONG) < 0) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong) { [...] It doesn't follow c99. Constants with explicit L prefix which doesn't fit in a ulong type will not be promoted to vlong but truncated as ulong. Removing '(c1 & Numlong) == 0' should do the trick, I think, but I don't like the code. The promotion of type from int to vlong only occurs just because all the compilers define int and long with the same width, and because that width is precisely 4 bytes. The rest of the original code is clearly not taking that fact into account which is in my opinion the correct approach. Forcing to specify explicitly the type if a different one from an int is desired is interesting, but the automatic conversion to an unsigned one already brakes the idea, doesn't it?. Some thoughts to help me decide what to do in my dist? Regards, adr. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-Mb38696f1b928920ec2c4d16d Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
[-- Attachment #1: Type: text/plain, Size: 838 bytes --] The short answer is whatever the historical reasons for the current interpretation(s), it should follow C99 rules to avoid confusion. One potential catch is that the compilers don't implement C99 rules when signed and unsigned values meet, but something closer to the original convention (which frankly was simpler) that the result (or comparison) is unsigned, portably, regardless of word size. I suspect things will break if that is changed, although again using the C99 rule is probably less confusing in the end, because both people and new incoming programs will expect the C99 interpretation. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M9c8aea84ddf712920e17bc0e Delivery options: https://9fans.topicbox.com/groups/9fans/subscription [-- Attachment #2: Type: text/html, Size: 1320 bytes --]
On Wed, 20 Apr 2022, Charles Forsyth wrote: > The short answer is whatever the historical reasons for the current interpretation(s), it > should follow C99 rules to avoid confusion. This should do the trick. I hope to not have to send another patch in half an hour... Charles, you are part of the p9f board, if you serve the 9legacy sources there it would be really easy to send you a patch. In my opinion it makes more sense to let the editions be archived as isos and substitute the sources 9p service with a 9legacy one. Just a suggestion, adr. /sys/src/cmd/cc/lex.c: [...] vv = yylval.vval; if(c1 & Numvlong || convvtox(vv, TUVLONG) > convvtox(vv, TULONG) || (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TLONG) < 0) { if((c1 & Numuns) || convvtox(vv, TVLONG) < 0) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong || convvtox(vv, TULONG) > convvtox(vv, TUINT) || (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TINT) < 0) { if((c1 & Numuns) || convvtox(vv, TLONG) < 0) { c = LULCONST; t = TULONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } if((c1 & Numuns) || convvtox(vv, TINT) < 0) { c = LUCONST; t = TUINT; goto nret; } c = LCONST; t = TINT; goto nret; nret: yylval.vval = convvtox(vv, t); if(yylval.vval != vv){ nearln = lineno; warn(Z, "truncated constant: %T %s", types[t], symb); } return c; [...] ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-Mabed3218d4cc85a6f48f03f7 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Quoth adr <adr@SDF.ORG>: > Some thoughts to help me decide what to do in my dist? The c99 rule is probably the one to follow, along side the plan9 unsigned preserving rule. When you have a patch, let me know -- I'll happily test and apply to 9front. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M4f89492d9145e3acd1d16df2 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
[-- Attachment #1: Type: text/plain, Size: 1946 bytes --] On Wed, 20 Apr 2022, ori@eigenstate.org wrote: > When you have a patch, let me know -- I'll happily test > and apply to 9front. Hi ori, this patch applyes to the sources served at 9front.org. By the way, do you plan to keep in sync http://only9fans.com/ori/git9/HEAD/info.html or should I forget about that repo? Regards, adr. --- /n/9front/sys/src/cmd/cc/lex.c Wed Apr 6 14:45:26 2022 +++ /tmp/lex.c Thu Apr 21 08:39:14 2022 @@ -848,16 +848,9 @@ yyerror("overflow in constant"); vv = yylval.vval; - /* - * c99 is silly: decimal constants stay signed, - * hex and octal go unsigned before widening. - */ - w = 32; - if((c1 & (Numdec|Numuns)) == Numdec) - w = 31; - if(c1 & Numvlong || (c1 & Numlong) == 0 && (uvlong)vv >= 1ULL<<w){ - if((c1&(Numdec|Numvlong)) == Numdec && vv < 1ULL<<32) - warn(Z, "int constant widened to vlong: %s", symb); + if(c1 & Numvlong || + convvtox(vv, TUVLONG) > convvtox(vv, TULONG) || + (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TLONG) < 0) { if((c1 & Numuns) || convvtox(vv, TVLONG) < 0) { c = LUVLCONST; t = TUVLONG; @@ -867,7 +860,9 @@ t = TVLONG; goto nret; } - if(c1 & Numlong) { + if(c1 & Numlong || + convvtox(vv, TULONG) > convvtox(vv, TUINT) || + (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TINT) < 0) { if((c1 & Numuns) || convvtox(vv, TLONG) < 0) { c = LULCONST; t = TULONG; ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-Mefe97ddcd27aad8f1a072269 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription [-- Warning: decoded text below may be mangled, UTF-8 assumed --] [-- Attachment #2: Type: text/x-diff; name=c99_int_constants_representation.diff, Size: 1086 bytes --] --- /n/9front/sys/src/cmd/cc/lex.c Wed Apr 6 14:45:26 2022 +++ /tmp/lex.c Thu Apr 21 08:39:14 2022 @@ -848,16 +848,9 @@ yyerror("overflow in constant"); vv = yylval.vval; - /* - * c99 is silly: decimal constants stay signed, - * hex and octal go unsigned before widening. - */ - w = 32; - if((c1 & (Numdec|Numuns)) == Numdec) - w = 31; - if(c1 & Numvlong || (c1 & Numlong) == 0 && (uvlong)vv >= 1ULL<<w){ - if((c1&(Numdec|Numvlong)) == Numdec && vv < 1ULL<<32) - warn(Z, "int constant widened to vlong: %s", symb); + if(c1 & Numvlong || + convvtox(vv, TUVLONG) > convvtox(vv, TULONG) || + (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TLONG) < 0) { if((c1 & Numuns) || convvtox(vv, TVLONG) < 0) { c = LUVLCONST; t = TUVLONG; @@ -867,7 +860,9 @@ t = TVLONG; goto nret; } - if(c1 & Numlong) { + if(c1 & Numlong || + convvtox(vv, TULONG) > convvtox(vv, TUINT) || + (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TINT) < 0) { if((c1 & Numuns) || convvtox(vv, TLONG) < 0) { c = LULCONST; t = TULONG;
Quoth adr <adr@SDF.ORG>: > On Wed, 20 Apr 2022, ori@eigenstate.org wrote: > > When you have a patch, let me know -- I'll happily test > > and apply to 9front. > > Hi ori, this patch applyes to the sources served at 9front.org. Highly appreciated -- I'm busy this week, but I'll take a look in the coming week. > By the way, do you plan to keep in sync > http://only9fans.com/ori/git9/HEAD/info.html or should I forget > about that repo? I'll update it. There are a few changes in flight that are being worked on, and once I commit them to 9front I'll sync back. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M355f4cd245105b2412ae537d Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Quoth adr <adr@SDF.ORG>: > > Hi ori, this patch applyes to the sources served at 9front.org. > By the way, do you plan to keep in sync > http://only9fans.com/ori/git9/HEAD/info.html or should I forget > about that repo? it's synced. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M5d091e2c50835bd999e39741 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Quoth adr <adr@SDF.ORG>: > On Wed, 20 Apr 2022, ori@eigenstate.org wrote: > > When you have a patch, let me know -- I'll happily test > > and apply to 9front. > > Hi ori, this patch applyes to the sources served at 9front.org. > By the way, do you plan to keep in sync > http://only9fans.com/ori/git9/HEAD/info.html or should I forget > about that repo? > > Regards, > adr. > > --- /n/9front/sys/src/cmd/cc/lex.c Wed Apr 6 14:45:26 2022 > +++ /tmp/lex.c Thu Apr 21 08:39:14 2022 > @@ -848,16 +848,9 @@ > yyerror("overflow in constant"); > > vv = yylval.vval; > - /* > - * c99 is silly: decimal constants stay signed, > - * hex and octal go unsigned before widening. > - */ > - w = 32; > - if((c1 & (Numdec|Numuns)) == Numdec) > - w = 31; > - if(c1 & Numvlong || (c1 & Numlong) == 0 && (uvlong)vv >= 1ULL<<w){ > - if((c1&(Numdec|Numvlong)) == Numdec && vv < 1ULL<<32) > - warn(Z, "int constant widened to vlong: %s", symb); > + if(c1 & Numvlong || > + convvtox(vv, TUVLONG) > convvtox(vv, TULONG) || > + (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TLONG) < 0) { > if((c1 & Numuns) || convvtox(vv, TVLONG) < 0) { > c = LUVLCONST; > t = TUVLONG; Thanks for the patch! I think this gets the condition slightly wrong. For a decimal number, we should promote: int -> long -> vlong. For a hex number, though, we want int -> uint -> long -> ulong -> vlong -> uvlong however with the condition above, 18446744073709551615 will get converted to a uvlong: c1 & Numvlong == 0: false -> no explicit conversion convvtox(vv, TUVLONG) > convvtox(vv, TULONG) true -> vv fits into a uvlong, but not a ulong, so we enter this. so then, we end up in this branch: if((c1 & Numuns) || convvtox(vv, TVLONG) < 0) { true, because convvtox(vv, TVLONG) < 0. I think we want: if(c1 & Numvlong || convvtox(vv, TUVLONG) != convvtox(vv, TULONG) || convvtox(vv, TLONG) < 0) { if((c1 & (Numdec|Numuns)) == 0 && ((c1 & Numuns) || convvtox(vv, TVLONG) < 0)) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong || convvtox(vv, TULONG) > convvtox(vv, TUINT) || convvtox(vv, TINT) < 0) { if((c1 & (Numdec|Numuns)) == 0 && ((c1 & Numuns) || convvtox(vv, TLONG) < 0)) { c = LULCONST; t = TULONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M4a92d1c5d54d32d3f8221d1e Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Quoth ori@eigenstate.org: > I think we want: I mean: if(c1 & Numvlong || convvtox(vv, TUVLONG) > convvtox(vv, TULONG) || (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TLONG) < 0) { if((c1 & (Numdec|Numuns)) == 0 && ((c1 & Numuns) || convvtox(vv, TVLONG) < 0)) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong || convvtox(vv, TULONG) > convvtox(vv, TUINT) || (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TINT) < 0) { if((c1 & (Numdec|Numuns)) == 0 && ((c1 & Numuns) || convvtox(vv, TLONG) < 0)) { c = LULCONST; t = TULONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M9b32e4c6660cf15f958c1e52 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
> it's synced. Nice. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-Mdbeaa9de34cf64c195becdec Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
> I think this gets the condition slightly wrong. For a decimal > number, we should promote: > > int -> long -> vlong. Yeah, i was thinking int -> long -> vlong -> uvlong when I wrote that, nice catch! ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-Mba95b4da9927e1dd42b446b3 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
On Sun, 1 May 2022, adr wrote: > Yeah, i was thinking > int -> long -> vlong -> uvlong No, it was a mistake, I forgot to check again that the conversion to unsigned when the constant is using the sign bit doesn't occurs to decimal constants. Also, reading 9front's mailing list I noticed the mess ignoring that convvtox will return vlong, type casting should do the trick: [...] vv = yylval.vval; if(c1 & Numvlong || (uvlong)convvtox(vv, TUVLONG) > convvtox(vv, TULONG) || (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TLONG) < 0) { if((c1 & Numuns) || convvtox(vv, TVLONG) < 0 && (c1 & Numdec) == 0) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong || convvtox(vv, TULONG) > convvtox(vv, TUINT) || (c1 & (Numdec|Numuns)) == Numdec && convvtox(vv, TINT) < 0) { if((c1 & Numuns) || convvtox(vv, TLONG) < 0 && (c1 & Numdec) == 0) { c = LULCONST; t = TULONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } if((c1 & Numuns) || convvtox(vv, TINT) < 0 && (c1 & Numdec) == 0) { c = LUCONST; t = TUINT; goto nret; } c = LCONST; t = TINT; goto nret; [...] It looks complex at first, but each line is telling you the reason to expand the constant, and when to use unsigned types first. For me is absolutely better than insert size assumptions in a piece of code which is making a great effort to let that to machine dependent code well kept in another place. I'll make some tests. adr ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M64323f2f5a4e8db8ea11250b Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
I think I got it right at last: [...] vv = yylval.vval; if(c1 & Numvlong || (uvlong)convvtox(vv, TUVLONG) > convvtox(vv, TULONG)){ /* unsigned suffix or hex occupying the sing bit */ if((c1 & Numuns) || convvtox(vv, TVLONG) < 0 && (c1 & Numdec) == 0) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong || (uvlong)convvtox(vv, TULONG) > convvtox(vv, TUINT)){ if((c1 & Numuns) || convvtox(vv, TLONG) < 0 && (c1 & Numdec) == 0) { c = LULCONST; t = TULONG; goto nret; } /* decimal occupying the sing bit */ if(convvtox(vv, TLONG) < 0 && (c1 & Numdec) == 0) { c = LVLCONST; t = TVLONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } if((c1 & Numuns) || convvtox(vv, TINT) < 0 && (c1 & Numdec) == 0) { c = LUCONST; t = TUINT; goto nret; } if(convvtox(vv, TINT) < 0 && (c1 & Numdec) == 0) { c = LLCONST; t = TLONG; goto nret; } c = LCONST; t = TINT; goto nret; [...] I used the cast also in (uvlong)convvtox(vv, TULONG) because the standard only specifies that long long can't be smaller than long. After playing with this I'm thinking about using the original code and substituting the warning for an error when a constant is truncated. Makes sense to me and Plan9's C isn't C99 anyway. adr. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M1f19d8edddfa4ec4ae0f5901 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Arrrg! [...] vv = yylval.vval; if(c1 & Numvlong || (uvlong)convvtox(vv, TUVLONG) > convvtox(vv, TULONG)){ /* unsigned suffix or hex occupying the sing bit */ if((c1 & Numuns) || convvtox(vv, TVLONG) < 0 && (c1 & Numdec) == 0) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong || (uvlong)convvtox(vv, TULONG) > convvtox(vv, TUINT)){ if((c1 & Numuns) || convvtox(vv, TLONG) < 0 && (c1 & Numdec) == 0) { c = LULCONST; t = TULONG; goto nret; } /* decimal occupying the sing bit */ if(convvtox(vv, TLONG) < 0 && (c1 & Numdec)) { c = LVLCONST; t = TVLONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } if((c1 & Numuns) || convvtox(vv, TINT) < 0 && (c1 & Numdec) == 0) { c = LUCONST; t = TUINT; goto nret; } if(convvtox(vv, TINT) < 0 && (c1 & Numdec)) { c = LLCONST; t = TLONG; goto nret; } c = LCONST; t = TINT; goto nret; [...] ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T22754f10b241991c-M489f3936d41217db381e8d09 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Building cmd I noticed that the case of sizeof(int) = sizeof(long) (which is the case in all compilers) wasn't managed with my changes. I wonder if this logic mess could be shortened without making impossible to read the 'if' conditions without throwing up (I've tried). I've compiled on arm the libraries, cmd and kernel (except ape) without any problem. adr. /sys/src/cmd/cc/lex.c [...] vv = yylval.vval; if(c1 & Numvlong || (uvlong)convvtox(vv, TUVLONG) > convvtox(vv, TULONG)){ /* unsigned suffix or hex occupying the sing bit */ if((c1 & Numuns) || convvtox(vv, TVLONG) < 0 && (c1 & Numdec) == 0) { c = LUVLCONST; t = TUVLONG; goto nret; } c = LVLCONST; t = TVLONG; goto nret; } if(c1 & Numlong || (uvlong)convvtox(vv, TULONG) > convvtox(vv, TUINT)){ if((c1 & Numuns) || convvtox(vv, TLONG) < 0 && (c1 & Numdec) == 0) { c = LULCONST; t = TULONG; goto nret; } /* decimal occupying the sing bit */ if(convvtox(vv, TLONG) < 0 && (c1 & Numdec)) { c = LVLCONST; t = TVLONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } if((c1 & Numuns) || convvtox(vv, TINT) < 0 && (c1 & Numdec) == 0) { c = LUCONST; t = TUINT; goto nret; } if(convvtox(vv, TINT) < 0 && (c1 & Numdec)) { /* size(int) = size(long) */ if(convvtox(vv, TLONG) < 0) { c = LVLCONST; t = TVLONG; goto nret; } c = LLCONST; t = TLONG; goto nret; } c = LCONST; t = TINT; goto nret; [...] ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T4bced580971a40a2-Mda3d0c619c734bb0e1cf708c Delivery options: https://9fans.topicbox.com/groups/9fans/subscription