December 2005 - Gpc

Inefficiency of `mod'
by Emil Jerabek 26 Oct '10

26 Oct '10

Hi, during some debugging I noticed that GPC generates very hairy code for the `mod' operator with signed arguments. A closer look revealed that the code contains a lot of branching on conditions which can be determined on compile time, and most surprisingly, the value of X mod Y is recomputed _twice_, if X is negative. (I've attached a commented assembler dump.) This may be partially a back-end problem (CSE?), but I feel there's something wrong in the relevant code in gpc-common.c (build_pascal_binary_op) as well. `exp1 mod exp2' gets translated into something like exp2 <= 0 ? error : exp1 >= 0 ? exp1 mod exp2 : ((-exp1) mod exp2) == 0 ? 0 : exp2 - (-exp1) mod exp2 This arranges that both arguments supplied to the "internal mod" are non-negative, but this information is not passed on -- the arguments are still declared as signed types, which apparently makes the back-end to produce the mess it does. I've done a little test: the following replacement for `mod', which differs from the built-in one essentially just by the type-casts to `Cardinal', inline operator xmod (X, Y: Integer) = Z: Integer; begin if Y <= 0 then RuntimeError (714); if X >= 0 then Z := Cardinal (X) mod Y else begin Z := Cardinal (-X) mod Y; if Z <> 0 then Z := Y - Z end end; indeed generates a better code, it measures about 15% faster for positive X and more than twice faster for negative X. YMMV. If this is possible in user code, it should work in the compiler internals as well :) The result type of the whole `mod' expression looks dubious too. IIUIC, `exp1 mod exp2' inherits the type of exp1. Now, the result of `mod' is guaranteed to be nonnegative and to fit in the type of exp2 (or more precisely, it is in 0 .. High (type of (exp2)) - 1), whereas it needn't have anything to do with the type of exp1: consider var X, Z: Integer; Y: LongInt; begin X := -8; Y := High (LongInt); Z := X mod Y { -> overflow } end Or am I missing something? Emil (BTW, where is fjf434c.pas?)

10 16

Building gpc-20051104 on i686-apple-darwin8
by Adriaan van Os 01 Mar '06

01 Mar '06

> [Darwin:gcc/p/test] adriaan% gpc -v > Reading specs from > /Developer/Pascal/gpc344x1/lib/gcc/i686-apple-darwin8/3.4.4/specs > Configured with: ../gcc-3.4.4/configure --enable-languages=c,pascal > --enable-threads=posix --target=i686-apple-darwin8 > --host=i686-apple-darwin8 --build=i686-apple-darwin8 > --prefix=/Developer/Pascal/gpc344x1 > Thread model: posix > gpc version 20051104, based on gcc-3.4.4 > > (more about the testsuite in the next message) Because of <http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22099> I ran the testsuite with --no-pic, so I had to tell the darwin linker to shut up complaining about that. [Darwin:gcc/p/test] adriaan% make EXTRA_PFLAGS="--no-pic -Wl,-read_only_relocs,suppress" rm -f *.dat *.o *.s *.i *.gpi *.gpd *.gpc core a.out stderr.out *.exe testmake.tmp dummy.c dummy.pas dummy.out diff_cr*.tmp fixcr fixcr.exe rm -f todo/a.out todo/*.exe todo/*.o todo/*.s todo/*.i todo/*.gpi todo/*.gpd todo/core GP= PC="gpc" PFLAGS=" --autobuild -g -O3 -W -Wall -Wno-unused --no-pic -Wl,-read_only_relocs,suppress " PFLAGS_NO_PATHS="-g -O3 -W -Wall -Wno-unused --no-pic -Wl,-read_only_relocs,suppress " SRCDIR="." TEST_MAKE_FLAG=test-make-flag "./test_run" "*.pas" | tee test_log | "./test_sum" -d Test Run By adriaan on 2005-11-14 00:37:07 Native configuration is i686-apple-darwin8 (Darwin.local) === gpc tests === Running target any Running testsuite ... UNSUPPORTED: agettext2test.pas UNSUPPORTED: agettexttest.pas UNSUPPORTED: aregextest.pas ^[[FAIL: backtracerr.pas ^R FAIL: bo4-19.pas FAIL: bprealtest.pas FAIL: emptyrec.pas FAIL: fay.pas UNSUPPORTED: fjf165a.pas FAIL: fjf35.pas FAIL: fjf464c.pas FAIL: fjf464e.pas FAIL: fjf908.pas FAIL: fproc.pas UNSUPPORTED: gmptest.pas FAIL: gpcu_c_u.pas FAIL: knuth1.pas FAIL: nicola4d.pas FAIL: nlgpp.pas FAIL: nlgpp2.pas FAIL: robert1a.pas FAIL: robert1b.pas FAIL: systemtest.pas FAIL: t6334-2.pas FAIL: t6p6p3p4.pas FAIL: waldek1.pas === gpc Summary === # of tests 4965 # of expected passes 4939 # of unexpected failures 21 # of unsupported tests 5 gpc version 20051104, based on gcc-3.4.4 Innocent ------- * systemtest failure is caused by a Max OS X bug * bo4-19.pas fails because of the linker option passed to the testsuite gpc: -read_only_relocs: linker input file unused because linking not done gpc: suppress: linker input file unused because linking not done, * emptyrec.pas, same * gpcu_c_u.pas, same * waldek1.pas, same Serious ------ * bprealtest.pas, segmentation fault Exception: EXC_BAD_ACCESS (0x0001) Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x00000041 Thread 0 Crashed: 0 dyld 0x8fe03d1b dyld::bindLazySymbol(mach_header const*, unsigned long*) + 119 1 dyld 0x8fe135cf stub_binding_helper_interface + 61 2 a.out 0x00001000 0x1000 + 0 3 a.out 0x000020c6 _p__M6_System_S63_Realtobpreal + 102 (system.pas:425) 4 a.out 0x0000305e _p__M0_main_program + 142 (bprealtest.pas:25) 5 a.out 0x000033ba main + 58 (<implicit code>:36) 6 a.out 0x00001cfe _start + 228 (crt.c:272) 7 a.out 0x00001c19 start + 41 ? maybe floating point exception * fay.pas, bus error Exception: EXC_BAD_ACCESS (0x0001) Codes: KERN_PROTECTION_FAILURE (0x0002) at 0xbffff6d0 Thread 0 Crashed: 0 <<00000000>> 0xbffff6d0 0 + -1073744176 1 a.out 0x00002da8 Q.5 + 168 (fay.pas:92) 2 a.out 0x00002dbd _p__M0_S6_Outer + 13 (fay.pas:90) 3 a.out 0x00002e2a main + 58 (<implicit code>:134) 4 a.out 0x0000211e _start + 228 (crt.c:272) 5 a.out 0x00002039 start + 41 ? procedural parameter and local procedure problems (? trampolines ?) the following test failures look the same: * backtraceerr.pas * fjf35.pas * fjf464c.pas * fjf464e.pas * fproc.pas, same * knuth1.pas, same * nicola4d.pas, same * nlgpp.pas * nlgpp2.pas * t6334-2.pas * t6p6p3p4.pas * fjf908.pas failed 89 1.000000000000000e+00 2.500000000000000e-01 * robert1a.pas x ** y = NaN trunc( x ** y) = 0 * robert1b.pas 10.0 ** y = NaN trunc( 10.0 ** y) = 0 Anything I can do to investigate this further and report it to gcc bugzilla ? Regards, Adriaan van Os

5 17

Re: GPC on i686-apple-darwin
by Waldek Hebisch 19 Jan '06

19 Jan '06

Adriaan van Os wrote: > You were quite right about -fPIC. All the failures, except emil9.pas > disappear with -fno-pic. However, many of the other tests fail with > -fno-pic, because the option disables dynamic linking on Mac OS X. The > PPC back-end has a workaround (-mdynamic-no-pic) but that doesn't seem > to work for i686-apple-darwin. > I have just realized that Darwin PIC problem have nothing to do with Linux PIC problems: Darwin does not use the binutils feature (ELF GOT relocations) that Linux uses. > Updating the linker and assembler to odcctools-20050327 had no effect > on testsuite results. > > So, I must either > - find a backport of i686-apple-darwin PIC fixes > - find a solution for gcc-3.4 bootstrap problems > - wait for gcc-4.0 support. > > Still, I am glad with the results so far. > Well, I was able build 3.3.6 based cross-compiler for i686-apple-darwin target, and I can see duplicate labels in the assembler output. It seems that 3.4.3 has changed the parts responsible for generating those labels. I have found that 3.4.[34] contain blatant bug with easy fix: --- gcc-3.4.4/gcc/config/i386/darwin.h.orig Thu Jun 16 01:10:29 2005 +++ gcc-3.4.4/gcc/config/i386/darwin.h Thu Jun 16 01:10:32 2005 @@ -43,7 +43,7 @@ /* Use the following macro for any Darwin/x86-specific command-line option translation. */ -#define SUBTARGET_OPTION_TRANSLATE_TABLE +#define SUBTARGET_OPTION_TRANSLATE_TABLE { 0, 0 } #define ASM_SPEC "-arch i386 \ %{Zforce_cpusubtype_ALL:-force_cpusubtype_ALL} \ After that I can build the compiler proper. However I get ICE when building gcc runtime. Trying gpc on knuth3.pas (to get assembler output) give me ICE. Both ICE-s seem related to generting labels needed in PIC code... So, it seems that nobody tried gcc-3.4.x for i686-darwin for almost a year (ChangeLog entry for SUBTARGET_OPTION_TRANSLATE_TABLE change is from 2004-08-23) -- Waldek Hebisch hebisch(a)math.uni.wroc.pl

2 9

Crash with sets under Solaris
by Pascal Viandier 16 Jan '06

16 Jan '06

Hi all, I am sorry to be back on this subject after a so long time, but I just hit the problem with a real-life program. I applied the patch you suggested and ran the gpc test suite again. It gives worse results than previously with 97 unexpected failures. If I run some programs under gdb I get Segmentation faults in various places in sets.pas. For example SetCopy (line 286) adam3g adam3h assumptions awindte2 awindtes dostest SetEqual (line 215) adam3k adam3l adam3m adam3n WordNumber (line 113) adam3q adam3r ancestor2 Test fail with no core dump az44 chief41 couper10 couper8 gpc gives: warning: `absolute' declaration increases required alignment and test fail chris4 chris5 Is there a patch to solve this problem? If you need more information, I will do my best to give it. Regards Pascal Viandier pascal(a)accovia.com -----Message d'origine----- De : gpc-owner(a)gnu.de [mailto:gpc-owner@gnu.de] De la part de Waldek Hebisch Envoyé : May 25, 2005 16:33 À : gpc(a)gnu.de Objet : Re: RE : bigsetdemo strange result Pascal Viandier wrote: > Hi, > > At your suggestion, I ran the gpc test suite. > Here is the summary result: > > Test Run By pascal on 2005-05-25 10:45:41 > Native configuration is sparc-sun-solaris2.6 (perfdev) > gpc 20050331, based on gcc-3.4.3, flags: -g -O3 -W -Wall -Wno-unused > > Nuber of tests: 4947 > Number of Ok: 4828 > Number of tests skipped: 7 > Number of failures: 56 (includes internal compiler error and some > coredumps) > Number of coredumps: 47 > Number of compiler errors: 2 > > > Should I worry? > Something is really broken. At first glance all failures involve sets, so it may be just a single problem. -- Waldek Hebisch hebisch(a)math.uni.wroc.pl

2 6

Success!
by Peter N Lewis 11 Jan '06

11 Jan '06

Thanks to all the help from Frank, Waldek, Adriaan, Gale and others, I now have all of Interarchy compiling and running under GPC. 270 files, 160,000 lines of code, 290 Mac Style object types, all compiling and working. There are probably a few glitches left here or there, but I've tested out the bulk of the program and nothing remains broken (that was not before - it is bug for bug compatible ;-). Thanks all! Peter. -- <http://www.stairways.com/> <http://download.stairways.com/>

6 5

variant record selector cannot be passed by reference
by prasad svr 04 Jan '06

04 Jan '06

Hi I am using gpc version 20050331, based on gcc-3.4.3 on solaris 10 intel platform. I have a record like below datobj_pointer = ^datobj_record; datobj_record = RECORD next : datobj_pointer; name : PASNameString; fct : Byte_integer; phystype: phystype_list; fix_ptr : datobj_pointer; nextfct : datobj_pointer; CASE subtype : all_obj_list OF Mea : (cur_mea : meaobj_pointer; com_mea : meaobj_pointer); Ind : (cur_ind : indobj_pointer; com_ind : indobj_pointer); Tim : (cur_tim : timobj_pointer; com_tim : timobj_pointer); Scr : (cur_scr : scrobj_pointer); Evt : (cur_evt : evtobj_pointer); Cmd : (cur_cmd : cmdobj_pointer); Con : (cur_con : conobj_pointer); Tbl : (cur_tbl : tblobj_pointer); Txf : (cur_txf : txfobj_pointer); Com : (cur_com : comobj_pointer; com_com : comobj_pointer); Par : (cur_par : param_pointer; com_par : param_pointer); Slv : (cur_slv : slave_pointer ); END; datobj : datobj_pointer; When I call the function like below r_w_phystype (iofil,i_o,buffer,20,datobj^.phystype,field); I am getting below error "variant record selector cannot be passed by reference". Our software is working on below platforms and compilers platform compiler sun sun pascal compiler vms vms pascal compiler digital unix digital unix pascal compiler windows borland delphi 7 In the above platforms we are not getting any problems,so please suggest whether I have any workaround? Regards Prasad -- _______________________________________________ Search for businesses by name, location, or phone number. -Lycos Yellow Pages http://r.lycos.com/r/yp_emailfooter/http://yellowpages.lycos.com/default.as…

3 3

Current distribution of gpc from www.cygwin.com does not work.
by Chris Hicks 04 Jan '06

04 Jan '06

I have just downloaded the test suite to try and resolve my problem with gpc that I posted earlier. The output from the test suite follows: $ make rm -f *.dat *.o *.s *.i *.gpi *.gpd *.gpc core a.out stderr.out *.exe testmake.tmp dummy.c dummy.pas dummy.out diff_cr*.tmp fixcr fixcr.exe rm -f todo/a.out todo/*.exe todo/*.o todo/*.s todo/*.i todo/*.gpi todo/*.gpd todo/core GP= PC="gpc" PFLAGS=" --autobuild -g -O3 -W -Wall -Wno-unused " PFLAGS_NO_PATHS="-g -O3 -W -Wall -Wno-unused " SRCDIR="." TEST_MAKE_FLAG=test-make-flag "./test_run" "*.pas" | tee test_log | "./test_sum" -d /usr/lib/gcc-lib/i686-pc-cygwin/3.3.3/../../../../i686-pc-cygwin/bin/ld: crtbegin.o: No such file: No such file or directory /usr/lib/gcc-lib/i686-pc-cygwin/3.3.3/../../../../i686-pc-cygwin/bin/ld: crtbegin.o: No such file: No such file or directory ./test_run: fatal: cannot compile a simple program Nothing was run. My problem seems to be due to the gpc/gcc distribution! Best regards, Chris.

4 5

MinGW version of GPC based on GCC 3.4.4?
by Gerrit P. Haase 02 Jan '06

02 Jan '06

Hello Prof, Is there a MinGW version of GPC based on GCC 3.4.4 available? Gerrit -- =^..^= Good Bye Johnny -------> http://www.ramones.com/

2 2

Unfriendly reset in gpc
by Brian Wichmann 29 Dec '05

29 Dec '05

If you do reset(t, 'myfile '); it fails due to the trailing space. If one is attempting to use something very close the ISO 7185, then removing the trailing spaces is virtually impossible. Of course, it is easy to do the following: procedure MyReset(var T: text; F: <some string type>); var i: integer; begin i := 1; while F[i] <> ' ' do i := i + 1; reset(T, F[1..i-1]); end; Hence I would like to suggest that reset removes trailing spaces, as on my old machine.... (ISO 7185, with very minimal additions). Brian Wichmann.

4 6

Re: gpc on mac
by Tom Schneider 29 Dec '05

29 Dec '05

Adriaan: > In that case, since there apparently is a linker on the machine and no > libSystemStubs.a, I assume that Xcode 1.x has been installed (maybe the > machine was updated later to Mac OS X 10.4). You can find out with "Get > Info" in the Mac OS X Finder for the file /Developer/Applications/Xcode > (or /Developer/Applications/Xcode.app). AH!!! Ok, I opened that and found from About that I have Xcode Version 1.1. Get Info gives that too and says it was modified Jan 23, 2004 5:13 AM. > Anyway, an update to Xcode 2.x > should solve the problem and install libSystemStubs.a. You can > download Xcode from Apple. The ADC membership for the purpose only > requires free registration, see > <http://developer.apple.com/membership/online.html>. Also, your copy of > Mac OS X system software may (or may not) have a third CD and Xcode is > on that third CD, which saves you a 800 MB download (or Xcode may be in > a special folder if the system software is on DVD). > > I would have to look if I can add libSystemStubs.a to the downloads > from my website for those who don't have Xcode installed or an older > version (if the license permits it). Even then, I don't know if it's > worth the trouble, because other problems may (or may not) arise when > Xcode tools are missing or not up-to-date. Ok, I am a member of ADC and could get Xcode 2.2 but it turns out that would be several hours of downloading, so I'm installing from my OS X 10.4 disk. This is more likely for people to do anyway. Ok, I have Xcode 2.0 now. Let's give it a whirl ... /Developer/Pascal/gpc345u2/bin/gpc test.p % a.out go to sleep!! BINGO! Thanks! It would help in your documentation to step people through checks for the installation (or can you put it into a script?) to make sure they have the Xcode they need. For whatever reason I didn't install the latest Xcode when I installed 10.4. Installation was pretty easy though from the DVD. Tom Dr. Thomas D. Schneider National Institutes of Health National Cancer Institute Center for Cancer Research Nanobiology Program Molecular Information Theory Group Frederick, Maryland 21702-1201 toms(a)ncifcrf.gov permanent email: toms(a)alum.mit.edu (use only if first address fails) http://www.ccrnp.ncifcrf.gov/~toms/ http://www.uwasa.fi/~ts/http/quote.html http://en.wikipedia.org/wiki/Top-posting

1 0