Switch statement optimization
Hi, while dealing with autogenerated code I found that GCC often outputs compare trees instead of branch tables for switch statements. I found that GCC uses the density of case labels within their spanned range as the key criterion. If the density of labels is smaller than 10% (33% for size optimization) then GCC uses compare instruction rather than a branch table. What I am wondering is why no real cost estimation is done here. In my case there are around 200 to 1500 labels within the 16K range, so they are completely resolved to compares. In contrast the Intel compiler outputs branch tables more likely. The question now is whether the factors 10 (3 for size opt) have been found using any standard benchmark. And shouldn't at least the factor be optimized for a standard benchmark (like the upcoming SPEC CPU 2006 INT) if this simple linear method is used for modeling a logarithmic behavior? Or should a cost estimation be introduced? For size optimization there could be a quite accurate architecture dependent size estimation. Opinions, Comments? Christian PS: Please CC me.
Re: Switch statement optimization
A paper at this year's GCC Summit talked about this: http://www.gccsummit.org/2006/view_abstract.php?content_key=18 You might like to follow up with Edmar (the author of the paper). Cheers, Ben
Questions regarding __register_frame_info
Hi
I have now everything in place for dynamically register the
debug frame information that my JIT (Just in time compiler)
generates.
I generate a CIE (Common information block), followed by
a series of FDE (Frame Description Entries) describing
each stack frame. The binary code is the same as gcc uses,
the contents of my stuff are identical to the contents of
the .eh_frame information.
There are several of those functions defined in unwind-dw2-fde.c:
__register_frame_info
__register_frame_info_bases
__register_frame_info_table_bases
If I use the
__register_frame_info
stuff, nothing happens and the program aborts.
Using __register_frame_jnfo_table_bases seems to
work better since it crashes a little bit further with a
hard crash.
Questions:
What is the procedure for registering the frame info?
I use following:
memset(&ob,0,sizeof(ob));
ob.pc_begin = (void *)-1;
ob.tbase = Parms.codebuf; // Machine instructions
ob.dbase = Parms.codebuf+myLccParms.text_size; // data of the
program
ob.s.i = 0;
ob.s.b.encoding = 0xff; //DW_EH_PE_omit;
__register_frame_info_table_bases(Parms.pUnwindTables,&ob,ob.tbase,ob.dbase);
"ob" is an object defined as follows:
struct object {
void *pc_begin;
void *tbase;
void *dbase;
union {
const struct dwarf_fde *single;
struct dwarf_fde **array;
struct fde_vector *sort;
} u;
union {
struct {
unsigned long sorted : 1;
unsigned long from_array : 1;
unsigned long mixed_encoding : 1;
unsigned long encoding : 8;
unsigned long count : 21;
} b;
size_t i;
} s;
#ifdef DWARF2_OBJECT_END_PTR_EXTENSION
char *fde_end;
#endif
struct object *next;
};
From the code of register_frame_info (in file unwind-dw2-fde.c)
that function just inserts the new data into a linked list, but it
does not do anything more. That is why probably it will never
work.
Could someone here explain me or tell me what to do exactly to register
the frame information?
This will be useful for all people that write JITs, for instance the
Java people,
and many others.
Thanks in advance for your help, and thanks for the help this group
has provided me already
jacob
Re: Switch statement optimization
In http://gcc.gnu.org/ml/gcc/2006-07/msg00156.html, your wrote: A paper at this year's GCC Summit talked about this: http://www.gccsummit.org/2006/view_abstract.php?content_key=18 You might like to follow up with Edmar (the author of the paper). But that was about optimizing the trees for an uneven probability distribution which could be found by profiling. This does not address the issue what to do when the distribution is mostly uniform or unknown. We could use a cheap hash and base start compare / branch trees in every hash bin. Say we have a 16 k range, 200 nodes, and want to keep the hash bin/node ratio between 2 and 4. Let i be the switch argument. Then we can calculate a 9 bit hash as (i ^ (x << n)) & 0x3fff, where n is a value between 5 and 9. We can pick the one which produces the flattest average search trees. Note that we no longer need to check that i is in range, as for ordinary switch dispatch tables. Moreover, on a target that can do effectively multi-way compares like the x86, in order to increase ILP, we can put into the table entry for each hash bin, in addition to the jump address, a value to compare i against before the dispatch jump takes place. If a cheap hash can be chosen so that there is no more than one entry per hash bin, we can even do a single compare in the dispatch code, go to the default case for non-match, and dispatch right to the handling code if we have a match.
Re: externs and thread local storage
On Fri, Jul 07, 2006 at 10:56:08PM -0400, Robert Dewar wrote: > That's not true, thread local storage can be done simply by mapping > hardware on some machines, where you swap maps on a context switch. Not with the semantics we have for __thread, which asserts that its address is valid in all threads. And, no one has tried this since Irix, wherein it didn't perform well. the effects of tlb flushing at context switch are heavy. r~
Re: request for new a syntactic design for C/C++.
"Vladimir 'Yu' Stepanov" <[EMAIL PROTECTED]> writes: > Current syntax C/C++: > > load_ptr = typeof(load_ptr)(((char *)init_ptr) - \ >offsetof(typeof(init_ptr), field); > > The offered syntax: > > &load_ptr->field = init_ptr; Interesting idea, but C/C++ programmers expect that an assignment sets the entire expression on the left of the '='. So I don't think this is a good syntax. In fact you can already write what you want with a macro, one quite similar to offsetof, so I don't think we need any new syntactic sugar here. > C++: > struct some_struct &link = other; This already means something in C++: it means to create a reference. Ian
Re: Questions regarding __register_frame_info
On Jul 10, 2006, at 6:57 AM, jacob navia wrote: What is the procedure for registering the frame info? If you don't get an answer, you may have to debug it. Just follow what something like eh6.C from the C++ testsuite does, and what it calls, when, and with what data, then mirror the eh6.C code in your language, JIT it, load it and then run it side by side with eh6.C and see if it behaves the same, or differently, and why. Be sure to use si/ni and display/1i $pc to see what is going on, if doing it by source lines doesn't reveal what is going on.
Mainline build broken (was: r115310 - in /trunk: fixincludes/ChangeLog fixi...)
On Mon, 2006-07-10 17:58:19 -, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote: > Author: lauras > Date: Mon Jul 10 17:58:18 2006 > New Revision: 115310 > > URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=115310 > Log: > fixincludes: > 2006-07-10 Laurynas Biveinis <[EMAIL PROTECTED]> > > PR bootstrap/20437 > * Makefile.in (configure, config.h.in): change into $(srcdir) > before autoconf or autoheader call. This patch broke building GCC because Makefile indention was done with spaces instead of a TAB. MfG, JBG -- Jan-Benedict Glaw [EMAIL PROTECTED]. +49-172-7608481 _ O _ "Eine Freie Meinung in einem Freien Kopf| Gegen Zensur | Gegen Krieg _ _ O für einen Freien Staat voll Freier Bürger" | im Internet! | im Irak! O O O ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA)); signature.asc Description: Digital signature
Re: Mainline build broken (was: r115310 - in /trunk: fixincludes/ChangeLog fixi...)
Will fix right now. 2006/7/10, Jan-Benedict Glaw <[EMAIL PROTECTED]>: On Mon, 2006-07-10 17:58:19 -, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote: > Author: lauras > Date: Mon Jul 10 17:58:18 2006 > New Revision: 115310 > > URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=115310 > Log: > fixincludes: > 2006-07-10 Laurynas Biveinis <[EMAIL PROTECTED]> > > PR bootstrap/20437 > * Makefile.in (configure, config.h.in): change into $(srcdir) > before autoconf or autoheader call. This patch broke building GCC because Makefile indention was done with spaces instead of a TAB. MfG, JBG -- Jan-Benedict Glaw [EMAIL PROTECTED]. +49-172-7608481 _ O _ "Eine Freie Meinung in einem Freien Kopf| Gegen Zensur | Gegen Krieg _ _ O für einen Freien Staat voll Freier Bürger" | im Internet! | im Irak! O O O ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA)); -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.1 (GNU/Linux) iD8DBQFEsqzIHb1edYOZ4bsRAoAzAKCI6dFjebFeBfbroJu1T4VraTSOFACfRJVx 8zcf06CJD/ZYwyxDi5APgUo= =RFuA -END PGP SIGNATURE- -- Laurynas
ICE in compute_frame_pointer_to_fb_displacement (building avr-gcc)
Hi, I was building a AVR cross compiler from current GCC trunk, and got an error. Build system was Ubuntu 6.06 on an Athlon, using the latest from CVS binutils but otherwise configured exactly as shown in http://www.nongnu.org/avr-libc/user-manual/install_tools.html The assertion failure that terminated the GCC build is appended. It's very easy to reproduce, but let me know if there's more info you need from me. I'm guessing that either not many folk have tried to build a GCC that supports the newest AVR chips, or those config+build instructions are wrong, since this is an error I saw with a few other recent GCCs in the last month or so. Anyone have a fix, workaround, or other suggestion? - Dave /home/tux/avr/gcc/obj-avr/./gcc/xgcc -B/home/tux/avr/gcc/obj-avr/./gcc/ -B/usr/local/avr/avr/bin/ -B/usr/local/avr/avr/lib/ -isystem /usr/local/avr/avr/include -isystem /usr/local/avr/avr/sys-include -O2 -O2 -g -O2 -DIN_GCC -DCROSS_COMPILE -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition -isystem ./include -DDF=SF -Dinhibit_libc -mcall-prologues -Os -g -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED -Dinhibit_libc -I. -I. -I../../gcc -I../../gcc/. -I../../gcc/../include -I../../gcc/../libcpp/include -I../../gcc/../libdecnumber -I../libdecnumber -DL_clear_cache -c ../../gcc/libgcc2.c -o libgcc/./_clear_cache.o ../../gcc/libgcc2.c: In function '__clear_cache': ../../gcc/libgcc2.c:1992: internal compiler error: in compute_frame_pointer_to_fb_displacement, at dwarf2out.c:10452 Please submit a full bug report, with preprocessed source if appropriate. See http://gcc.gnu.org/bugs.html> for instructions. make[3]: *** [libgcc/./_clear_cache.o] Error 1 make[3]: Leaving directory `/home/tux/avr/gcc/obj-avr/gcc' make[2]: *** [stmp-multilib] Error 2 make[2]: Leaving directory `/home/tux/avr/gcc/obj-avr/gcc' make[1]: *** [all-gcc] Error 2 make[1]: Leaving directory `/home/tux/avr/gcc/obj-avr' make: *** [all] Error 2
Re: request for new a syntactic design for C/C++.
On Jul 9, 2006, at 10:52 PM, Vladimir 'Yu' Stepanov wrote: I would like to offer one expansion for C/C++. Did you just reinvent downcasting in C++? If so, C++ already has that feature!? As for C, C, I'd claim C already has that feature[1], you merely have to put in a #define into your library to make it look pretty. In C, since this is a library issue, the compiler would be the wrong place to put the feature. Since it can be done with a macro, there is little need to change the grammar of the language. 1 - Contentious, should be decided by C committee IMHO.
Re: Mainline build broken (was: r115310 - in /trunk: fixincludes/ChangeLog fixi...)
This patch broke building GCC because Makefile indention was done with spaces instead of a TAB. Obvious fix commited, r115313. That will teach me how to think "oh well that's a tiny patch I sent a month ago, I'll just copy it from the mail archives instead of locating it on my disk". Sorry. -- Laurynas
Re: gcc breakage
You forgot a ChangeLog entry for r115313 | lauras | 2006-07-10 12:44:48 -0700 (Mon, 10 Jul 2006) | 2 lines Fix spaces to tabs in the last commit. I'm not sure. I think that my ChangeLog entry by my previous commit (the breaking one) covers this change too, because if I had managed to commit it correctly the first time, repository would be in the same state as it is now. Do such changes require ChangeLog entries? Should I commit it now? -- Laurynas
Re: gcc breakage
On Mon, Jul 10, 2006 at 11:24:55PM +0300, Laurynas Biveinis wrote: > >You forgot a ChangeLog entry for > > > > > >r115313 | lauras | 2006-07-10 12:44:48 -0700 (Mon, 10 Jul 2006) | 2 lines > > > >Fix spaces to tabs in the last commit. > > I'm not sure. I think that my ChangeLog entry by my previous commit > (the breaking one) covers this change too, because if I had managed to > commit it correctly the first time, repository would be in the same > state as it is now. > > Do such changes require ChangeLog entries? Should I commit it now? > It's my understanding that every change requires a ChangeLog entry. There is now no consistent record of the difference between 115310 and 115311 without actually looking at "svn log fixincludes/Makefile.in". Perhaps, a quick question on IRC of the correct protocol is appropriate? -- Steve
Re: gcc breakage
On Mon, Jul 10, 2006 at 01:34:12PM -0700, Steve Kargl wrote: > It's my understanding that every change requires a ChangeLog > entry. There is now no consistent record of the difference > between 115310 and 115311 without actually looking at "svn log > fixincludes/Makefile.in". > > Perhaps, a quick question on IRC of the correct protocol is > appropriate? It is fairly typical to omit the changelog entry in "fix my last patch" commits. Also in "fix the ChangeLog" commits. -- Daniel Jacobowitz CodeSourcery
/bin/sh: build/genmodes: No such file or directory (was Re: gcc-4.2-20060708 is now available)
> Snapshot gcc-4.2-20060708 is now available on > ftp://gcc.gnu.org/pub/gcc/snapshots/4.2-20060708/ something is brkoen here ... [~/gcc] >../gcc-src/gcc-4.2-20060708/gcc/configure checking build system type... i686-pc-linux-gnu checking host system type... i686-pc-linux-gnu checking target system type... i686-pc-linux-gnu checking LIBRARY_PATH variable... ok checking GCC_EXEC_PREFIX variable... ok checking whether to place generated files in the source directory... no checking whether a default linker was specified... no checking whether a default assembler was specified... no checking for gcc... gcc checking for C compiler default output file name... a.out checking whether the C compiler works... yes checking whether we are cross compiling... no checking for suffix of executables... checking for suffix of object files... o checking whether we are using the GNU C compiler... yes checking whether gcc accepts -g... yes checking for gcc option to accept ANSI C... none needed checking whether gcc and cc understand -c and -o together... yes checking how to run the C preprocessor... gcc -E checking for inline... inline checking for egrep... grep -E checking for ANSI C header files... yes checking for sys/types.h... yes checking for sys/stat.h... yes checking for stdlib.h... yes checking for string.h... yes checking for memory.h... yes checking for strings.h... yes checking for inttypes.h... yes checking for stdint.h... yes checking for unistd.h... yes checking for void *... yes checking size of void *... 4 checking for short... yes checking size of short... 2 checking for int... yes checking size of int... 4 checking for long... yes checking size of long... 4 checking for long long... yes checking for long long... (cached) yes checking size of long long... 8 checking for __int64... no checking whether gcc accepts -Wno-long-long... yes checking whether gcc accepts -Wno-variadic-macros... no checking whether gcc accepts -Wno-overlength-strings... no checking whether gcc accepts -Wold-style-definition... yes checking whether gcc accepts -Wmissing-format-attribute... yes checking whether gcc accepts -Wc++-compat... no checking valgrind.h usability... no checking valgrind.h presence... no checking for valgrind.h... no checking whether make sets $(MAKE)... yes checking for gawk... gawk checking whether ln -s works... yes checking whether ln works... yes checking for ranlib... ranlib checking for a BSD compatible install... /usr/bin/install -c checking for cmp's capabilities... gnucompare checking for mktemp... yes checking for makeinfo... makeinfo checking for modern makeinfo... yes checking for recent Pod::Man... yes checking for flex... flex checking for bison... bison checking for nm... nm checking for ar... ar checking for GNU C library... yes checking for ANSI C header files... (cached) yes checking whether time.h and sys/time.h may both be included... yes checking whether string.h and strings.h may both be included... yes checking for sys/wait.h that is POSIX.1 compatible... yes checking for limits.h... yes checking for stddef.h... yes checking for string.h... (cached) yes checking for strings.h... (cached) yes checking for stdlib.h... (cached) yes checking for time.h... yes checking for iconv.h... yes checking for fcntl.h... yes checking for unistd.h... (cached) yes checking for sys/file.h... yes checking for sys/time.h... yes checking for sys/mman.h... yes checking for sys/resource.h... yes checking for sys/param.h... yes checking for sys/times.h... yes checking for sys/stat.h... (cached) yes checking for direct.h... no checking for malloc.h... yes checking for langinfo.h... yes checking for ldfcn.h... no checking for locale.h... yes checking for wchar.h... yes checking for thread.h... no checking for pthread.h... yes checking for CHAR_BIT... yes checking whether byte ordering is bigendian... no checking for collect2 libraries... none required checking for library containing exc_resume... no checking for library containing ldexp... none required checking for inttypes.h... yes checking for times... yes checking for clock... yes checking for kill... yes checking for getrlimit... yes checking for setrlimit... yes checking for atoll... yes checking for atoq... no checking for sysconf... yes checking for strsignal... yes checking for getrusage... yes checking for nl_langinfo... yes checking for scandir... yes checking for alphasort... yes checking for gettimeofday... yes checking for mbstowcs... yes checking for wcswidth... yes checking for mmap... yes checking for mincore... yes checking for setlocale... yes checking for clearerr_unlocked... yes checking for feof_unlocked... yes checking for ferror_unlocked... yes checking for fflush_unlocked... yes checking for fgetc_unlocked... yes checking for fgets_unlocked... yes checking for fileno_unlocked... yes checking for fprintf_unlocked... no checking for fputc_unlocked... yes checking for fputs_unlocked... yes checking for fread_unlocked... yes checking for fwrite_
Re: gcc breakage
On Jul 10, 2006, at 1:24 PM, Laurynas Biveinis wrote: Do such changes require ChangeLog entries? For my own beyond trivial fixes that I catch quickly (say, less than 24 hours), normally I just fix them with a `fix typo' type svn log entry. When in doubt, or if you're fixing someone else's work, or someone else filed a PR for it already, do up a changelog entry. If other people notice, I'd be tempted to say, do up a changelog entry.
Re: ICE in compute_frame_pointer_to_fb_displacement (building avr-gcc)
David Brownell wrote: /home/tux/avr/gcc/obj-avr/./gcc/xgcc -B/home/tux/avr/gcc/obj-avr/./gcc/ -B/usr/local/avr/avr/bin/ -B/usr/local/avr/avr/lib/ -isystem /usr/local/avr/avr/include -isystem /usr/local/avr/avr/sys-include -O2 -O2 -g -O2 -DIN_GCC -DCROSS_COMPILE -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition -isystem ./include -DDF=SF -Dinhibit_libc -mcall-prologues -Os -g -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED -Dinhibit_libc -I. -I. -I../../gcc -I../../gcc/. -I../../gcc/../include -I../../gcc/../libcpp/include -I../../gcc/../libdecnumber -I../libdecnumber -DL_clear_cache -c ../../gcc/libgcc2.c -o libgcc/./_clear_cache.o .../../gcc/libgcc2.c: In function '__clear_cache': .../../gcc/libgcc2.c:1992: internal compiler error: in compute_frame_pointer_to_fb_displacement, at dwarf2out.c:10452 Looks like bug 26504. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26504 -- Jim Wilson, GNU Tools Support, http://www.specifix.com
Re: ICE in compute_frame_pointer_to_fb_displacement (building avr-gcc)
On Jul 11, 2006, at 4:49 AM, David Brownell wrote: Hi, Anyone have a fix, workaround, or other suggestion? Yes don't use --with-dwarf2 :). This is PR 26504: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26504 Also someone should try the patch in PR 26015: http://gcc.gnu.org/bugzilla/show_bug.cgi?id= 26015: Which most likely will also fix this. -- Pinski
Running aclocal in libstdc++-v3 directory
I am trying to make a configure change in the libstdc++-v3 directory but when I run aclocal (even on an unmodified libstdc++-v3 directory from the top-of-tree), I get an error message. Does anyone else see this? [hpsje] $ aclocal aclocal:configure.ac:85: warning: macro `AM_PROG_LIBTOOL' not found in library configure.ac:138: warning: AC_PROG_LD is m4_require'd but is not m4_defun'd configure.ac:138: AC_PROG_LD is required by... acinclude.m4:307: GLIBCXX_CHECK_LINKER_FEATURES is expanded from... configure.ac:138: the top level [hpsje] $ aclocal --version aclocal (GNU automake) 1.9.6 Written by Tom Tromey <[EMAIL PROTECTED]> Copyright (C) 2005 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
Re: Running aclocal in libstdc++-v3 directory
On Mon, Jul 10, 2006 at 03:14:36PM -0700, Steve Ellcey wrote: > > I am trying to make a configure change in the libstdc++-v3 directory but > when I run aclocal (even on an unmodified libstdc++-v3 directory from > the top-of-tree), I get an error message. Does anyone else see this? My rule of thumb: Check the options in automake's generated Makefile.in. You probably need a -I path. -- Daniel Jacobowitz CodeSourcery
Re: Running aclocal in libstdc++-v3 directory
> My rule of thumb: Check the options in automake's generated > Makefile.in. > > You probably need a -I path. > > -- > Daniel Jacobowitz > CodeSourcery Ah, I see "ACLOCAL_AMFLAGS = -I . -I .. -I ../config", if I add those -I options to my aclocal call I get no errors or warnings when running aclocal. Thanks for the tip. Steve Ellcey [EMAIL PROTECTED]
Re: Running aclocal in libstdc++-v3 directory
> I am trying to make a configure change in the libstdc++-v3 directory but > when I run aclocal (even on an unmodified libstdc++-v3 directory from > the top-of-tree), I get an error message. Does anyone else see this? The current fashion for regenerating the config/make bits is to just run: autoreconf at the top level libstdc++-v3 source directory. -benjamin
[PATCH] Wcoercion: removed overflow check
This patch is an improved version over the one proposed in
http://gcc.gnu.org/ml/gcc-patches/2006-07/msg00098.html
It doesn't introduce any features. The main difference is the removal
of TREE_CONSTANT_OVERFLOW check from real_isinteger function,
following the new policy about not handling overflow in the front end.
This patch also fixes some minor issues with whitespace and comments.
Thanks to Roger Sayle for his helpful comments.
Bootstrapped and tested in trunk revision 115112.
As the original one, it is composed by three patches:
wcoercion-1of3-split.patch : splits current functionality of
Wconversion in two different options and it modifies testcases
accordingly. (this one hasn't changed at all, it is included here for
completeness)
wcoercion-2of3-real_isinteger-2.patch : adds a new function which
detects when a real value can be exactly represented as an integer.
wcoercion-3of3-coercion_warning-2.patch : adds the new functionality and
testcases.
diff -pEbaur --unidirectional-new-file --exclude='*svn*' --exclude='*~' pristine/gcc/c-common.c wcoercion/gcc/c-common.c
--- pristine/gcc/c-common.c 2006-07-01 13:37:45.0 +0100
+++ wcoercion/gcc/c-common.c 2006-07-01 13:40:44.0 +0100
@@ -949,7 +949,7 @@ overflow_warning (tree value)
}
/* Print a warning if a large constant is truncated to unsigned,
- or if -Wconversion is used and a constant < 0 is converted to unsigned.
+ or if -Wcoercion is used and a constant < 0 is converted to unsigned.
Invoke this function on every expression that might be implicitly
converted to an unsigned type. */
@@ -969,7 +969,7 @@ unsigned_conversion_warning (tree result
warning (OPT_Woverflow,
"large integer implicitly truncated to unsigned type");
else
- warning (OPT_Wconversion,
+ warning (OPT_Wcoercion,
"negative integer implicitly converted to unsigned type");
}
}
diff -pEbaur --unidirectional-new-file --exclude='*svn*' --exclude='*~' pristine/gcc/c.opt wcoercion/gcc/c.opt
--- pristine/gcc/c.opt 2006-06-04 09:53:03.0 +0100
+++ wcoercion/gcc/c.opt 2006-06-25 20:33:21.0 +0100
@@ -141,6 +141,10 @@ Wchar-subscripts
C ObjC C++ ObjC++ Var(warn_char_subscripts)
Warn about subscripts whose type is \"char\"
+Wcoercion
+C ObjC C++ ObjC++ Var(warn_coercion)
+Warn for implicit type conversions that may change a value
+
Wcomment
C ObjC C++ ObjC++
Warn about possibly nested block comments, and C++ comments spanning more than one physical line
@@ -150,8 +154,12 @@ C ObjC C++ ObjC++
Synonym for -Wcomment
Wconversion
-C ObjC C++ ObjC++ Var(warn_conversion)
-Warn about possibly confusing type conversions
+C++ ObjC++ Var(warn_conversion)
+Undocumented
+
+Wtraditional-conversion
+C ObjC Var(warn_traditional_conversion)
+Warn of prototypes causing type conversions different from what would happen in the absence of prototype
Wctor-dtor-privacy
C++ ObjC++ Var(warn_ctor_dtor_privacy)
diff -pEbaur --unidirectional-new-file --exclude='*svn*' --exclude='*~' pristine/gcc/c-typeck.c wcoercion/gcc/c-typeck.c
--- pristine/gcc/c-typeck.c 2006-07-01 13:38:05.0 +0100
+++ wcoercion/gcc/c-typeck.c 2006-07-01 13:40:43.0 +0100
@@ -2389,7 +2389,7 @@ convert_arguments (tree typelist, tree v
{
/* Optionally warn about conversions that
differ from the default conversions. */
- if (warn_conversion || warn_traditional)
+ if (warn_traditional_conversion || warn_traditional)
{
unsigned int formal_prec = TYPE_PRECISION (type);
@@ -2465,8 +2465,8 @@ convert_arguments (tree typelist, tree v
}
/* Detect integer changing in width or signedness.
These warnings are only activated with
- -Wconversion, not with -Wtraditional. */
- else if (warn_conversion && INTEGRAL_TYPE_P (type)
+ -Wtraditional-conversion, not with -Wtraditional. */
+ else if (warn_traditional_conversion && INTEGRAL_TYPE_P (type)
&& INTEGRAL_TYPE_P (TREE_TYPE (val)))
{
tree would_have_been = default_conversion (val);
@@ -2479,7 +2479,7 @@ convert_arguments (tree typelist, tree v
and the actual arg is that enum type. */
;
else if (formal_prec != TYPE_PRECISION (type1))
- warning (OPT_Wconversion, "passing argument %d of %qE "
+ warning (OPT_Wtraditional_conversion, "passing argument %d of %qE "
"with different width due to prototype",
argnum, rname);
else if (TYPE_UNSIGNED (type) == TYPE_UNSIGNED (type1))
@@ -2502,11 +2502,11 @@ convert_arguments (tree typelist, tree v
&& TYPE_UNSIGNED (TREE_TYPE (val)))
;
else if (TYPE_UNSIGNED (type))
- warning (OPT_Wconversion, "passing argument %d of %qE "
+ warning (OPT_Wtraditional_conversion, "passing argument %d of %qE "
"as unsigned due to prototype",
argnum, rname);
else
- warning (OPT_Wconversion, "passing argument %d of %qE "
+ warning (OPT_Wtraditional_conversion, "passing argument %d of %qE "
"as signed
Re: make proto fails
Andreas Jaeger wrote: # Fix for argument ‘clean_text_p’ might be clobbered by ‘longjmp’ or ‘vfork’ unprotoize.o-warn = -Wno-error This is OK. PR 21161 and PR 24239 are different instances of the same problem, so we don't need another bug report for this. The previously mentioned PR 21059 is a different issue. -- Jim Wilson, GNU Tools Support, http://www.specifix.com
Someone broke darwin?
Gosh, so close... I found that the below patch gets me one step closer to it building. I'm sure it is wrong... but it should be enough of a hint for the right person to know how to fix it. Doing diffs in libstdc++-v3: --- libstdc++-v3/include/ext/codecvt_specializations.h.~1~ 2005-12-20 12:50:02.0 -0800 +++ libstdc++-v3/include/ext/codecvt_specializations.h 2006-07-10 18:03:56.0 -0700 @@ -42,6 +42,8 @@ // Define this here so codecvt.cc can have _S_max_size definition. #define _GLIBCXX_USE_ENCODING_STATE 1 +#include + _GLIBCXX_BEGIN_NAMESPACE(__gnu_cxx) /// @brief Extension to use icov for dealing with character encodings. -- The next problem after that one is that: /Volumes/mrs3/net/gcc-darwin/./gcc/xgcc -shared-libgcc -B/Volumes/ mrs3/net/gcc-darwin/./gcc -nostdinc++ -L/Volumes/mrs3/net/gcc-darwin/ powerpc-apple-darwin8.5.0/libstdc++-v3/src -L/Volumes/mrs3/net/gcc- darwin/powerpc-apple-darwin8.5.0/libstdc++-v3/src/.libs -B/Volumes/ mrs3/Packages/gcc-20060518/powerpc-apple-darwin8.5.0/bin/ -B/Volumes/ mrs3/Packages/gcc-20060518/powerpc-apple-darwin8.5.0/lib/ -isystem / Volumes/mrs3/Packages/gcc-20060518/powerpc-apple-darwin8.5.0/include - isystem /Volumes/mrs3/Packages/gcc-20060518/powerpc-apple-darwin8.5.0/ sys-include -Winvalid-pch -Wno-deprecated -x c++-header -g -O2 -I/ Volumes/mrs3/net/gcc-darwin/powerpc-apple-darwin8.5.0/libstdc++-v3/ include/powerpc-apple-darwin8.5.0 -I/Volumes/mrs3/net/gcc-darwin/ powerpc-apple-darwin8.5.0/libstdc++-v3/include -I/Volumes/mrs3/net/ gcc/libstdc++-v3/libsupc++ -O2 -g /Volumes/mrs3/net/gcc/libstdc++-v3/ include/precompiled/extc++.h -o powerpc-apple-darwin8.5.0/bits/extc+ +.h.gch/O2g.gch /Volumes/mrs3/net/gcc-darwin/powerpc-apple-darwin8.5.0/libstdc++-v3/ include/ext/pb_ds/detail/binary_heap_/binary_heap_.hpp:235: internal compiler error: Bus error Please submit a full bug report, with preprocessed source if appropriate. See http://gcc.gnu.org/bugs.html> for instructions. make[4]: *** [powerpc-apple-darwin8.5.0/bits/extc++.h.gch/O2g.gch] Error 1 die... :-( Not yet sure why...
Re: /bin/sh: build/genmodes: No such file or directory (was Re: gcc-4.2-20060708 is now available)
On Monday 10 July 2006 22:42, [EMAIL PROTECTED] wrote: > > Snapshot gcc-4.2-20060708 is now available on > > ftp://gcc.gnu.org/pub/gcc/snapshots/4.2-20060708/ > > something is brkoen here ... > [~/gcc] > > >../gcc-src/gcc-4.2-20060708/gcc/configure > ups, should of course have been ../gcc-src/gcc-4.2-20060708/configure
Re: ICE in compute_frame_pointer_to_fb_displacement (building avr-gcc)
> > Anyone have a fix, workaround, or other suggestion? > > Yes don't use --with-dwarf2 :). Only a short term workaround... ;) > This is PR 26504: > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26504 Comment #8 there seems on-the-mark, so the PR 26015 fix has two half-confirmations now: > Also someone should try the patch in PR 26015: > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26015: Which assumes the frame pointer is in a register (pair) on AVR ... seems normally true (r29:r28 == Y), but maybe someone more AVR-savvy should verify there are no nasty corner cases. > Which most likely will also fix this. Comment #8 said that the AVR version of that VAX patch helped -- and I got past that in the build, see my AVR SVN patch attached -- but there was a second issue with libssp that he worked around with "configure ... --disable-libssp". I took a quick look at the libssp thing, and it looks like maybe it's being built too early ... the linker needs some chip-specfic runtime (crts8515.o) that won't exist until AVR-libc gets built -- well after gcc gets built, that is. So, I have a successful build but haven't yet tried to flash and run the code yet. I'll let you know later if this compiler actually _works_ ... thanks! - Dave "make" in object directory gave: Checking multilib configuration for libssp... mkdir -p -- avr/libssp Configuring in avr/libssp configure: creating cache ./config.cache checking build system type... i686-pc-linux-gnu checking host system type... avr-unknown-none checking target system type... avr-unknown-none checking for a BSD-compatible install... /usr/bin/install -c checking whether build environment is sane... yes checking for gawk... gawk checking whether make sets $(MAKE)... yes checking for avr-strip... /usr/local/avr/avr/bin/strip checking for --enable-version-specific-runtime-libs... no checking whether to enable maintainer-specific portions of Makefiles... no checking for avr-gcc... /home/tux/avr/gcc/obj-avr/./gcc/xgcc -B/home/tux/avr/gcc/obj-avr/./gcc/ -B/usr/local/avr/avr/bin/ -B/usr/local/avr/avr/lib/ -isystem /usr/local/avr/avr/include -isystem /usr/local/avr/avr/sys-include checking for C compiler default output file name... configure: error: C compiler cannot create executables See `config.log' for more details. make[1]: *** [configure-target-libssp] Error 1 make[1]: Leaving directory `/home/tux/avr/gcc/obj-avr' make: *** [all] Error 2 Command exited with non-zero status 2 avr/libssp/config.log held: configure:2307: checking for C compiler default output file name configure:2310: /home/tux/avr/gcc/obj-avr/./gcc/xgcc -B/home/tux/avr/gcc/obj-avr/./gcc/ -B/usr/local/avr/avr/bin/ -B/usr/local/avr/avr/lib/ -isystem /usr/local/avr/avr/include -isystem /usr/local/avr/avr/sys-include -O2 -g -O2conftest.c >&5 /usr/local/avr/avr/bin/ld: crts8515.o: No such file: No such file or directory configure:2313: $? = 1 configure: failed program was: | /* elided, it's perfectly legit and not the problem */ configure:2352: error: C compiler cannot create executables This gets GCC to build on AVR-8 "--with-dwarf2"; a "--disable-libssp" is necessary too, and should maybe be the default. --- gcc/config/avr/avr.h (revision 115307) +++ gcc/config/avr/avr.h (working copy) @@ -271,6 +271,8 @@ #define FIRST_PARM_OFFSET(FUNDECL) 0 +#define FRAME_POINTER_CFA_OFFSET(FUNDECL) 0 + #define STACK_BOUNDARY 8 #define STACK_POINTER_REGNUM 32
Re: Someone broke darwin?
On Jul 10, 2006, at 6:20 PM, Mike Stump wrote:
The next problem after that one is that:
ext/pb_ds/detail/binary_heap_/binary_heap_.hpp:235: internal
compiler error: Bus error
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html> for instructions.
make[4]: *** [powerpc-apple-darwin8.5.0/bits/extc++.h.gch/O2g.gch]
Error 1
Curious, not caused by changes in gcc/cp since 113399, so the
regression appears to be a libstdc++ issue. Anyway, I've tracked the
issue down to mangling the type for pch, and with this change:
Doing diffs in .:
--- ./mangle.c.~1~ 2006-07-10 18:56:25.0 -0700
+++ ./mangle.c 2006-07-10 19:16:52.0 -0700
@@ -2306,7 +2306,8 @@ write_template_arg (tree node)
/* A template appearing as a template arg is a template
template arg. */
write_template_template_arg (node);
else if ((TREE_CODE_CLASS (code) == tcc_constant && code !=
PTRMEM_CST)
- || (abi_version_at_least (2) && code == CONST_DECL))
+ || (abi_version_at_least (2) && code == CONST_DECL
+ && TREE_CODE (DECL_INITIAL (node)) != SCOPE_REF))
write_template_arg_literal (node);
else if (DECL_P (node))
{
--
it then encodes:
(gdb) pt
local bindings <0x2e72438>> value simple_value>> context
heap_>
chain >
readonly constant invariant private VOID file /Volumes/mrs3/net/
gcc-darwin/powerpc-apple-darwin8.5.0/libstdc++-v3/inc\
lude/ext/pb_ds/detail/binary_heap_/binary_heap_.hpp line 117
align 1 context binary_heap_> initial
chain >
arg 0 type_0 type_5 type_6
VOID
align 8 symtab 0 alias set -1 context 0x2d83930 detail>
no-binfo use_template=1 interface-unknown
chain >>
arg 1
local bindings <0x0>>>
as something like:
LZNS5_12simple_valueEE
I've not had a chance to think about it, so don't know if that is
right or not. Someone else care to let me know if I'm on the right
track?
If so, someone wanna approve it for mainline to get us bootstrapping
again?
Re: [PATCH] Wcoercion: removed overflow check
> This patch is an improved version over the one proposed in Please avoid cross-posting. This should go on [EMAIL PROTECTED] only. -- Eric Botcazou
