Re: gcc compile error
"David O'Brien" wrote: On Wed, Dec 29, 1999 at 11:37:18AM -0800, Amancio Hasty wrote: This is the scoop. ..snip.. gcc -v Using builtin specs. gcc version 2.95.2 19991024 (release) ..snip.. Without -O or -O2 the program compiles okay. What other platforms w/gcc 2.95 have you tried to build this X11 version on? What other non-Linux platforms w/gcc 2.95 have you tried to build this X11 version on? Do you have a problem if you download the official GCC 2.95.2 release and build it the old GNU'fashion way? Does a purely stock GCC 2.95 bomb on this?? If so, it is a GCC problem and the issue should be raised with Cygnus. Yes it is a gcc-2.95.2 bug. It's fixed in the 2.96 snapshot in ports/lang/gcc-devel. Cheers, -Peter -- Peter Wemm - [EMAIL PROTECTED]; [EMAIL PROTECTED]; [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
"David O'Brien" wrote:
On Wed, Dec 29, 1999 at 11:37:18AM -0800, Amancio Hasty wrote:
This is the scoop.
..snip..
gcc -v
Using builtin specs.
gcc version 2.95.2 19991024 (release)
..snip..
Without -O or -O2 the program compiles okay.
What other platforms w/gcc 2.95 have you tried to build this X11 version
on? What other non-Linux platforms w/gcc 2.95 have you tried to build
this X11 version on? Do you have a problem if you download the official
GCC 2.95.2 release and build it the old GNU'fashion way? Does a purely
stock GCC 2.95 bomb on this?? If so, it is a GCC problem and the issue
should be raised with Cygnus.
Yes it is a gcc-2.95.2 bug. It's fixed in the 2.96 snapshot in
ports/lang/gcc-devel.
Yeap,
gcc-devel installed from the ports directory:
gcc -v
Reading specs from /usr/local/lib/gcc-lib/i386-portbld-freebsd4.0/2.96/specs
gcc version 2.96 19991110 (experimental)
{root} gcc -c -O2 bug.c
That is as Peter said the bug is fixed on the 2.96 snapshot and I am currently
rebuilding X on my other system will find out more when the build finishes in
about 1/2 hour.
Take Care
--
Amancio Hasty
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Amancio Hasty wrote: I really doubt that I am the only one here that can get XFree86 3.9.xxx -curr ent . Nevertheless if it can help out to fix the default compiler here is the infor mation which you reguested. Command Line executed to generate the output file and with the exception of t he gcc option "-E" it is the exact same command which failed to compile xf86vmod e.c The current official release (ports/lang/egcs) also has this problem. However, ports/lang/gcc-devel (the current snapshot) compiles this file OK. Looking at the changelog, it seems that there has been a *lot* of work on the reload/clobber/etc area. So, if you want to compile the current XFree86, it looks like it'll need the port. The bad news is that the cygnus folks are absolutely determined to not support any dual-format (a.out and elf) stuff for us, this means that you won't be able to build a.out libraries. Cheers, -Peter To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Tue, 28 Dec 1999, David O'Brien wrote:
Perhaps a later snapshot of gcc will work .
GCC 2.95.2 is a *RELEASED* version. We don't use snapshots as the base
compiler. What every the problem is 4.0 will live with it unless someone
narrows down the problem more.
The latest Polygraph (ver 2.2.8) proxy-cache benchmarking source
(Available at http://www.ircache.net/Polygraph/) does not
compile on my very recent 4.0 system. (rebuilding world now...)
I get this error, after dl'ing polygraph-2.2.8-src.tar.gz, running
'configure', then running 'make':
...
c++ -g -O3 -Wall -Wwrite-strings -Woverloaded-virtual -Iinclude
-Ixstd/include -Ipgl/include -Ixparser/include -I.. -DHAVE_CONFIG_H -c
Connection.cc
In file included from include/PortMgr.h:29,
from Connection.cc:33:
include/LevelStat.h:55: invalid type `const char[1]' for default argument
to `const String '
*** Error code 1
Stop in /usr/home/andyf/src/polygraph/src.
*** Error code 1
Stop in /usr/home/andyf/src/polygraph.
The "offending" code looks like this:
ostream print(ostream os, const String pfx = "") const;
When I did a 'touch Connection.o ; make' it compiled another source file
that included PortMgr.h / LevelStat.h just fine, but bombed out elsewhere!
--
-- David([EMAIL PROTECTED])
--
:{ [EMAIL PROTECTED]
Andy Farkas
System Administrator
Speednet Communications
http://www.speednet.com.au/
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
XFree86 3.9.xxx was cvsup on December 24th I am sorry but this is sufficient information to reproduce the problem. Not if you actually want the problem solved. There's this thing called "making it easy on the people you're demanding things of" in order that they might have some chance of actually doing it. Insisting that someone cvsup the entire X source tree, as you've clearly *already* done, hardly falls into the category of attempting to save the gcc maintainers work in attempting to track down the problem you're complaining about. - Jordan To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Wed, 29 Dec 1999, Jordan K. Hubbard wrote: XFree86 3.9.xxx was cvsup on December 24th I am sorry but this is sufficient information to reproduce the problem. [snip] Insisting that someone cvsup the entire X source tree, as you've clearly *already* done, hardly falls into the category of attempting to save the gcc maintainers work in attempting to track down the problem you're complaining about. Hmmm... are "regular people" even allowed access to the XFree86 cvs server? I thought that was just reserved for XFree86 developers? Don't tell me we have to become XFree86 developers just to solve a simple gcc problem. :) BTW, I hope they release 3.9.17 soon - Donn To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Sending out an attachment of that size to a public mailing list was hardly necessary, and the increasing stridency of your posts leading up to this only serve to indicate that you may be heading in the truly wrong direction with all this and seriously need to rethink your strategy before you do something that has people howling for your blood. - Jordan To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
: : FreeBSD -current was last cvsup on my system on Dec 23 18:59. : : XFree86 3.9.xxx was cvsup on December 24th I am sorry but this is : sufficient information to reproduce the problem. : :No, it is not, because we can't *get* XFree86 3.9.xxx -current, so we :cannot reproduce it, can we? : :You either need to give us enough to build that part, or provide a :preprocessed file so we can feed the backends directly, or fix it yourself. Sure you can well, you can get close anyway. ftp ftp.xfree86.org cd snapshots get 3.9.16.tar.gz I've been keeping up to date myself hoping that the I128 driver (for the SGI flatscreen) would support the I^2 protocol the SGI uses to control backlight brightness, there being no external knob on the monitor to fiddle with. -Matt To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Sending out an attachment of that size to a public mailing list was
hardly necessary, and the increasing stridency of your posts leading
up to this only serve to indicate that you may be heading in the truly
wrong direction with all this and seriously need to rethink your
strategy before you do something that has people howling for your
blood.
- Jordan
Okay folks,
This is the scoop.
For the benefit of those which don't have access to the latest XFree86 release you can
download the file from ftp://rah.star-gate.com/pub/bug.c
uname -a
FreeBSD x.star-gate.com 4.0-CURRENT FreeBSD 4.0-CURRENT #2: Mon Dec 27 13:55:25 PST
1999 [EMAIL PROTECTED]:/usr/src/sys/compile/MUADIB i386
gcc -v
Using builtin specs.
gcc version 2.95.2 19991024 (release)
gcc -c -O2 bug.c
xf86vmode.c: In function `ProcXF86VidModeGetMonitor':
xf86vmode.c:1320: Unable to generate reloads for:
(insn 300 298 302 (parallel[
(set (reg:SI 0 %eax)
(fix:SI (fix:SF (subreg:SF (reg:SI 0 %eax) 0
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -34 [0xffde])) 0))
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -36 [0xffdc])) 0))
(clobber (mem:SI (plus:SI (reg:SI 6 %ebp)
(const_int -40 [0xffd8])) 0))
(clobber (scratch:HI))
] ) 145 {fix_truncsfsi2+1} (insn_list 293 (nil))
(expr_list:REG_DEAD (reg:SI 0 %eax)
(expr_list:REG_UNUSED (scratch:HI)
(nil
--
Without -O or -O2 the program compiles okay.
gcc -c bug.c
According to Peter, the ports/lang/gcc-devel (the current snapshot) appears to
compile this program fine -- that is the compiler bug has been fixed.
So there appears to be two solutions to get around this problem:
1. compile without -O
or
2. installed the latest snapshot of gcc.
Take Care Guys
--
Amancio Hasty
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Wed, 29 Dec 1999, Amancio Hasty wrote: ! !Without -O or -O2 the program compiles okay. ! !gcc -c bug.c ! Ouch! This looks an awful lot like the last report with `GCC' and `problem' in the subject. As Matt just mentionned one or two posts ago, and as I observed in the last thread, have you tried making some of the variables `volatile?' Assuming of course that the code does compile well without the optimization flags, one would assume that the "problem" occurs because of caching of certain variable values in registers (not that this should be a problem, mind you; one would assume that the optimization isn't performed too well). Due to lack of time, and generally speaking, lack of experience with GCC, I haven't taken a detailed shot at debugging this. Bosko. -- Bosko Milekic Email: [EMAIL PROTECTED] WWW:http://pages.infinit.net/bmilekic/ -- To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Hi, We just have a buggy version of gcc and it appears that the register allocator is the main problematic area. This is not really a problem for me because what I first try out is a newer version of gcc if I can not get around the compile problem. At any rate remember that at least my reported compiler bug appears to be fixed in the latest gcc-devel port. On Wed, 29 Dec 1999, Amancio Hasty wrote: ! !Without -O or -O2 the program compiles okay. ! !gcc -c bug.c ! Ouch! This looks an awful lot like the last report with `GCC' and `problem' in the subject. As Matt just mentionned one or two posts ago, and as I observed in the last thread, have you tried making some of the variables `volatile?' Assuming of course that the code does compile well without the optimization flags, one would assume that the "problem" occurs because of caching of certain variable values in registers (not that this should be a problem, mind you; one would assume that the optimization isn't performed too well). Due to lack of time, and generally speaking, lack of experience with GCC, I haven't taken a detailed shot at debugging this. Bosko. -- Bosko Milekic Email: [EMAIL PROTECTED] WWW:http://pages.infinit.net/bmilekic/ -- -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Thu, Dec 30, 1999 at 02:40:46AM +1100, Andy Farkas wrote: In file included from include/PortMgr.h:29, from Connection.cc:33: include/LevelStat.h:55: invalid type `const char[1]' for default argument to `const String ' ..snip.. The "offending" code looks like this: ostream print(ostream os, const String pfx = "") const; What is the prototype for this function? Hint take a look. You have bad (ie, now invalid by stricter compilers) C++ code. When I did a 'touch Connection.o ; make' it compiled another source file that included PortMgr.h / LevelStat.h just fine, but bombed out elsewhere! I can't even fathom what you expected to accomplish by this. Are you a programmer? -- -- David([EMAIL PROTECTED]) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Wed, Dec 29, 1999 at 11:37:18AM -0800, Amancio Hasty wrote: This is the scoop. ..snip.. gcc -v Using builtin specs. gcc version 2.95.2 19991024 (release) ..snip.. Without -O or -O2 the program compiles okay. What other platforms w/gcc 2.95 have you tried to build this X11 version on? What other non-Linux platforms w/gcc 2.95 have you tried to build this X11 version on? Do you have a problem if you download the official GCC 2.95.2 release and build it the old GNU'fashion way? Does a purely stock GCC 2.95 bomb on this?? If so, it is a GCC problem and the issue should be raised with Cygnus. So there appears to be two solutions to get around this problem: ..snip.. 3. Raise this issue with Cygnus. -- -- David([EMAIL PROTECTED]) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
3. Raise this issue with Cygnus. Not really Cygnus is the wrong organization to raise this issue . As someone else pointed out the gcc-devel port does not exhibit the bug which I posted. -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
3. Raise this issue with Cygnus. Not really Cygnus is the wrong organization to raise this issue . Could you *please* explain why??? Gcc 2.96 will not be out before 4.0. So Gcc 2.95.x is what is going into 4.0. Now should a Gcc 2.95.3 were to come out, then we'd get a new compiler for 4.0. As someone else pointed out the gcc-devel port does not exhibit the bug which I posted. Lets think about this in FreeBSD terms -- 4.0 does not have some problem that 3.4-R does. However it wasn't known that 3.4-R had this problem and the reason why 4.0 does not have this problem is due to *tons* of code changes, not necessarily in response to this problem. Now would the FreeBSD Project like to know that there is a problem with 3.4-R that isn't in 4.0 along with a test case to show the problem? Would the FreeBSD Project, maybe just maybe be interested in fixing the problem and releasing a 3.5-R? Just like FreeBSD, EGCS has two code branches. -- -- David([EMAIL PROTECTED]) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Thu, Dec 30, 1999 at 02:21:48PM +1100, Andy Farkas wrote: ...the idea was to continue the make process further along to where another source file that also included LevelStat.h got compiled, to check whether it bombs as well - it didn't. ``make -k'' might have been a better choice as you don't have to remember to rm the fake .o later. -- -- David([EMAIL PROTECTED]) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Gcc 2.96 will not be out before 4.0. So Gcc 2.95.x is what is going into 4.0. Now should a Gcc 2.95.3 were to come out, then we'd get a new compiler for 4.0. Lets think about this in FreeBSD terms -- 4.0 does not have some problem that 3.4-R does. However it wasn't known that 3.4-R had this problem and Hi, I am running FreeBSD -current and I got no clue as to what is going on FreeBSD 3.xxx. The problem was reported for FreeBSD -current. Okay, lets hope that gcc 2.9.53 comes out before the release of FreeBSD 4.0. Take Care -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Wed, Dec 29, 1999 at 07:43:07PM -0800, Amancio Hasty wrote: Gcc 2.96 will not be out before 4.0. So Gcc 2.95.x is what is going into 4.0. Now should a Gcc 2.95.3 were to come out, then we'd get a new compiler for 4.0. Lets think about this in FreeBSD terms -- 4.0 does not have some problem that 3.4-R does. However it wasn't known that 3.4-R had this problem and I am running FreeBSD -current and I got no clue as to what is I *REALIZE* THAT. I was making an _analogy_. The problem was reported for FreeBSD -current. The problem most likely needs to be reported for GCC 2.95, *NOT* FreeBSD. Okay, lets hope that gcc 2.9.53 comes out before the release of FreeBSD 4.0. Not unless someone like you gets involved and files a bug report with Cygnus. -- -- David([EMAIL PROTECTED]) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Hi David, Last time, the problem does not exist with the gcc-devel port which directly implies that the problem has been fixed so I see no point on reporting the bug to Cygnus. I can in the future report the bug to Cygnus if the bug has not been fixed in a subsequent snapshot. I will play with gcc-devel some tonite to see if I can use it for all my ports. Some background: I am working on XFree86 to help bring about yuv + scaling hardware support -- my main priority is X for I if get too side tracked I may miss my window of opportunity to hack . Take Care -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Actually, that's one of the newest versions of gcc. Perhaps a later snapshot of gcc will work . GCC 2.95.2 is a *RELEASED* version. We don't use snapshots as the base compiler. What every the problem is 4.0 will live with it unless someone narrows down the problem more. -- -- David([EMAIL PROTECTED]) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Sun, Dec 26, 1999 at 12:13:42PM -0500, Donn Miller wrote: I get similar errors trying to compile aview from ports. I just updated my ports tree, so that can't be the problem. See the attached make.log. There's something about not allowing access to the cx register. ... sstring.h:493: fixed or forbidden register 2 (cx) was spilled for class CREG. These are easy to fix. Please use `send-pr' to submit a PR and hopefully I can get to it. -- -- David([EMAIL PROTECTED]) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Thats nice . Now we have a compiler which fails to build X. Actually, that's one of the newest versions of gcc. Perhaps a later snapshot of gcc will work . GCC 2.95.2 is a *RELEASED* version. We don't use snapshots as the base compiler. What every the problem is 4.0 will live with it unless someone narrows down the problem more. -- -- David([EMAIL PROTECTED]) -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Thats nice . Now we have a compiler which fails to build X. This seems like hyperbole. I'm able to build X just fine with the -current compiler, so to directly imply that we can't do so flies in the face of common sense and experience. - Jordan To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Yes, I can rightfully claim that you cannot build XFree86 3.9 xxx dated December 24 with the default gcc compiler on FreeBSD -current. I am running FreeBSD -current and XFree86 3.9xxx current. Thats nice . Now we have a compiler which fails to build X. This seems like hyperbole. I'm able to build X just fine with the -current compiler, so to directly imply that we can't do so flies in the face of common sense and experience. - Jordan -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Tue, 28 Dec 1999, Amancio Hasty wrote: Yes, I can rightfully claim that you cannot build XFree86 3.9 xxx dated December 24 with the default gcc compiler on FreeBSD -current. I am running FreeBSD -current and XFree86 3.9xxx current. Typically this empty space would contain the output of the build failure, more details, or even a patch. You might want to find out why your mailer is thrashing it. Thats nice . Now we have a compiler which fails to build X. This seems like hyperbole. I'm able to build X just fine with the -current compiler, so to directly imply that we can't do so flies in the face of common sense and experience. -- - bill fumerola - [EMAIL PROTECTED] - BF1560 - computer horizons corp - - ph:(800) 252-2421 - [EMAIL PROTECTED] - [EMAIL PROTECTED] - To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
The details were previously posted by . Good Nite -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Tue, 28 Dec 1999, Amancio Hasty wrote: The details were previously posted by . Good Nite Whatever ate the details on your previous post seems to be at it again. -- - bill fumerola - [EMAIL PROTECTED] - BF1560 - computer horizons corp - - ph:(800) 252-2421 - [EMAIL PROTECTED] - [EMAIL PROTECTED] - To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
On Tue, 28 Dec 1999, Amancio Hasty wrote:
The details were previously posted by .
Good Nite
Whatever ate the details on your previous post seems to be at it again.
Here is a copy of my mail post which I received from the -current mail list
that is my post went out intact.
Posted on Sat, 25 Dec 1999 23:17:45 -0800
---
While trying to compile the latest version of XFree86 I found the following
compiler bug.
{root} gcc -v
Using builtin specs.
gcc version 2.95.2 19991024 (release)
I assume that this an old version of gcc...
xc/programs/Xserver/Xext
cc -c -pipe -O2 -ansi -pedantic -Dasm=__asm -Wall -Wpointer-arith -Wstrict-prototypes
-Wmissing-prototypes -Wmissing-declarations -Wredundant-decls -Wnested-externs -I.
-I../include -I../../../exports/include/X11 -I../../../include/extensions
-I../../../include/fonts -I../mi -I../../../programs/Xserver/hw/xfree86/common
-I../../../include/fonts -I../../.. -I../../../exports/include -DCSRG_BASED -DSHAPE
-DXINPUT -DXKB -DLBX -DXAPPGROUP -DXCSECURITY -DTOGCUP -DXF86BIGFONT -DDPMSExtension
-DPIXPRIV -DPANORAMIX -DGCCUSESGAS -DAVOID_GLYPHBLT -DPIXPRIV -DSINGLEDEPTH
-DXFreeXDGA -DXvExtension -DXFree86LOADER -DXFree86Server -DXF86VIDMODE
-DX_BYTE_ORDER=X_LITTLE_ENDIAN -DSMART_SCHEDULE -DNDEBUG -DFUNCPROTO=15
-DNARROWPROTO xf86vmode.c
In file included from xf86vmode.c:41:
../include/extnsionst.h:91: warning: function declaration isn't a prototype
../include/extnsionst.h:100: warning: function declaration isn't a prototype
../include/extnsionst.h:106: warning: function declaration isn't a prototype
xf86vmode.c: In function `ProcXF86VidModeGetMonitor':
xf86vmode.c:1320: Unable to generate reloads for:
(insn 300 298 302 (parallel[
(set (reg:SI 0 %eax)
(fix:SI (fix:SF (subreg:SF (reg:SI 0 %eax) 0
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -34 [0xffde])) 0))
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -36 [0xffdc])) 0))
(clobber (mem:SI (plus:SI (reg:SI 6 %ebp)
(const_int -40 [0xffd8])) 0))
(clobber (scratch:HI))
] ) 145 {fix_truncsfsi2+1} (insn_list 293 (nil))
(expr_list:REG_DEAD (reg:SI 0 %eax)
(expr_list:REG_UNUSED (scratch:HI)
(nil
*** Error code 1
--
Amancio Hasty
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Amancio Hasty wrote:
On Tue, 28 Dec 1999, Amancio Hasty wrote:
The details were previously posted by .
Good Nite
Whatever ate the details on your previous post seems to be at it again.
Here is a copy of my mail post which I received from the -current mail list
that is my post went out intact.
It isn't much use to anyone like this... We need to be able to reproduce
this. What release of xfree86 is this? 3.9.16 as released? if not you will
need to give us code if you want help.
Failing that, how about a cc -E of the file in question so we can compile it
standalone and see what is going wrong with the inline asm?
Posted on Sat, 25 Dec 1999 23:17:45 -0800
---
While trying to compile the latest version of XFree86 I found the following
compiler bug.
{root} gcc -v
Using builtin specs.
gcc version 2.95.2 19991024 (release)
I assume that this an old version of gcc...
xc/programs/Xserver/Xext
cc -c -pipe -O2 -ansi -pedantic -Dasm=__asm -Wall -Wpointer-arith -Wstrict-pr
ototypes -Wmissing-prototypes -Wmissing-declarations -Wredundant-decls -Wne
sted-externs -I.
-I../include -I../../../exports/include/X11 -I../../../include/extensions
-I../../../include/fonts -I../mi -I../../../programs/Xserver/hw/xfree86/
common
-I../../../include/fonts -I../../.. -I../../../exports/include -DCSRG_BASED
-DSHAPE -DXINPUT -DXKB -DLBX -DXAPPGROUP -DXCSECURITY -DTOGCUP -DXF86BIGF
ONT -DDPMSExtension
-DPIXPRIV -DPANORAMIX -DGCCUSESGAS -DAVOID_GLYPHBLT -DPIXPRIV -DSINGLEDEPTH
-DXFreeXDGA -DXvExtension -DXFree86LOADER -DXFree86Server -DXF86VIDMODE
-DX_BYTE_ORDER=X_LITTLE_ENDIAN -DSMART_SCHEDULE -DNDEBUG -DFUNCPROTO=15 -DN
ARROWPROTO xf86vmode.c
In file included from xf86vmode.c:41:
../include/extnsionst.h:91: warning: function declaration isn't a prototype
../include/extnsionst.h:100: warning: function declaration isn't a prototype
../include/extnsionst.h:106: warning: function declaration isn't a prototype
xf86vmode.c: In function `ProcXF86VidModeGetMonitor':
xf86vmode.c:1320: Unable to generate reloads for:
(insn 300 298 302 (parallel[
(set (reg:SI 0 %eax)
(fix:SI (fix:SF (subreg:SF (reg:SI 0 %eax) 0
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -34 [0xffde])) 0))
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -36 [0xffdc])) 0))
(clobber (mem:SI (plus:SI (reg:SI 6 %ebp)
(const_int -40 [0xffd8])) 0))
(clobber (scratch:HI))
] ) 145 {fix_truncsfsi2+1} (insn_list 293 (nil))
(expr_list:REG_DEAD (reg:SI 0 %eax)
(expr_list:REG_UNUSED (scratch:HI)
(nil
*** Error code 1
--
Amancio Hasty
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Cheers,
-Peter
--
Peter Wemm - [EMAIL PROTECTED]; [EMAIL PROTECTED]; [EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Look I stated previously and only a few hours ago:
---
Yes, I can rightfully claim that you cannot build XFree86 3.9 xxx dated
December 24 with the default gcc compiler on FreeBSD -current.
I am running FreeBSD -current and XFree86 3.9xxx current.
---
More precisely :
FreeBSD -current was last cvsup on my system on Dec 23 18:59.
XFree86 3.9.xxx was cvsup on December 24th I am sorry but this is
sufficient information to reproduce the problem.
Amancio Hasty wrote:
On Tue, 28 Dec 1999, Amancio Hasty wrote:
The details were previously posted by .
Good Nite
Whatever ate the details on your previous post seems to be at it again.
Here is a copy of my mail post which I received from the -current mail list
that is my post went out intact.
It isn't much use to anyone like this... We need to be able to reproduce
this. What release of xfree86 is this? 3.9.16 as released? if not you will
need to give us code if you want help.
Failing that, how about a cc -E of the file in question so we can compile it
standalone and see what is going wrong with the inline asm?
Posted on Sat, 25 Dec 1999 23:17:45 -0800
---
While trying to compile the latest version of XFree86 I found the following
compiler bug.
{root} gcc -v
Using builtin specs.
gcc version 2.95.2 19991024 (release)
I assume that this an old version of gcc...
xc/programs/Xserver/Xext
cc -c -pipe -O2 -ansi -pedantic -Dasm=__asm -Wall -Wpointer-arith -Wstrict-pr
ototypes -Wmissing-prototypes -Wmissing-declarations -Wredundant-decls -Wne
sted-externs -I.
-I../include -I../../../exports/include/X11 -I../../../include/extensions
-I../../../include/fonts -I../mi -I../../../programs/Xserver/hw/xfree86/
common
-I../../../include/fonts -I../../.. -I../../../exports/include -DCSRG_BASED
-DSHAPE -DXINPUT -DXKB -DLBX -DXAPPGROUP -DXCSECURITY -DTOGCUP -DXF86BIGF
ONT -DDPMSExtension
-DPIXPRIV -DPANORAMIX -DGCCUSESGAS -DAVOID_GLYPHBLT -DPIXPRIV -DSINGLEDEPTH
-DXFreeXDGA -DXvExtension -DXFree86LOADER -DXFree86Server -DXF86VIDMODE
-DX_BYTE_ORDER=X_LITTLE_ENDIAN -DSMART_SCHEDULE -DNDEBUG -DFUNCPROTO=15 -DN
ARROWPROTO xf86vmode.c
In file included from xf86vmode.c:41:
../include/extnsionst.h:91: warning: function declaration isn't a prototype
../include/extnsionst.h:100: warning: function declaration isn't a prototype
../include/extnsionst.h:106: warning: function declaration isn't a prototype
xf86vmode.c: In function `ProcXF86VidModeGetMonitor':
xf86vmode.c:1320: Unable to generate reloads for:
(insn 300 298 302 (parallel[
(set (reg:SI 0 %eax)
(fix:SI (fix:SF (subreg:SF (reg:SI 0 %eax) 0
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -34 [0xffde])) 0))
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -36 [0xffdc])) 0))
(clobber (mem:SI (plus:SI (reg:SI 6 %ebp)
(const_int -40 [0xffd8])) 0))
(clobber (scratch:HI))
] ) 145 {fix_truncsfsi2+1} (insn_list 293 (nil))
(expr_list:REG_DEAD (reg:SI 0 %eax)
(expr_list:REG_UNUSED (scratch:HI)
(nil
*** Error code 1
--
Amancio Hasty
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Cheers,
-Peter
--
Peter Wemm - [EMAIL PROTECTED]; [EMAIL PROTECTED]; [EMAIL PROTECTED]
--
Amancio Hasty
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Amancio Hasty wrote: Look I stated previously and only a few hours ago: --- Yes, I can rightfully claim that you cannot build XFree86 3.9 xxx dated December 24 with the default gcc compiler on FreeBSD -current. I am running FreeBSD -current and XFree86 3.9xxx current. --- More precisely : FreeBSD -current was last cvsup on my system on Dec 23 18:59. XFree86 3.9.xxx was cvsup on December 24th I am sorry but this is sufficient information to reproduce the problem. No, it is not, because we can't *get* XFree86 3.9.xxx -current, so we cannot reproduce it, can we? You either need to give us enough to build that part, or provide a preprocessed file so we can feed the backends directly, or fix it yourself. Cheers, -Peter To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: gcc compile error
Amancio Hasty wrote:
While trying to compile the latest version of XFree86 I found the following
compiler bug.
{root} gcc -v
Using builtin specs.
gcc version 2.95.2 19991024 (release)
I assume that this an old version of gcc...
Actually, that's one of the newest versions of gcc.
-snip-
xf86vmode.c: In function `ProcXF86VidModeGetMonitor':
xf86vmode.c:1320: Unable to generate reloads for:
(insn 300 298 302 (parallel[
(set (reg:SI 0 %eax)
(fix:SI (fix:SF (subreg:SF (reg:SI 0 %eax) 0
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -34 [0xffde])) 0))
(clobber (mem:HI (plus:SI (reg:SI 6 %ebp)
(const_int -36 [0xffdc])) 0))
(clobber (mem:SI (plus:SI (reg:SI 6 %ebp)
(const_int -40 [0xffd8])) 0))
(clobber (scratch:HI))
] ) 145 {fix_truncsfsi2+1} (insn_list 293 (nil))
(expr_list:REG_DEAD (reg:SI 0 %eax)
(expr_list:REG_UNUSED (scratch:HI)
(nil
*** Error code 1
I get similar errors trying to compile aview from ports. I just
updated my ports tree, so that can't be the problem. See the
attached make.log. There's something about not allowing access
to the cx register.
- Donn
=== Building for aview-1.2
cc -O -pipe -I/usr/X11R6/include -I/usr/local/include -c flip.c
flip.c: In function `dcd_delta_flc':
sstring.h:493: Invalid `asm' statement:
sstring.h:493: fixed or forbidden register 2 (cx) was spilled for class CREG.
flip.c: In function `dcd_delta_fli':
sstring.h:493: Invalid `asm' statement:
sstring.h:493: fixed or forbidden register 2 (cx) was spilled for class CREG.
flip.c: In function `dcd_byte_run':
sstring.h:493: Invalid `asm' statement:
sstring.h:493: fixed or forbidden register 2 (cx) was spilled for class CREG.
flip.c: In function `dcd_black':
sstring.h:669: Invalid `asm' statement:
sstring.h:669: fixed or forbidden register 2 (cx) was spilled for class CREG.
flip.c: In function `dcd_literal':
sstring.h:493: Invalid `asm' statement:
sstring.h:493: fixed or forbidden register 2 (cx) was spilled for class CREG.
*** Error code 1
Stop in /usr/ports/graphics/aview/work/aview-1.2.
*** Error code 1
Stop in /usr/ports/graphics/aview.
*** Error code 1
Stop in /usr/ports/graphics/aview.
*** Error code 1
Stop in /usr/ports/graphics/aview.
Re: gcc compile error
Actually, that's one of the newest versions of gcc. Perhaps a later snapshot of gcc will work . attached make.log. There's something about not allowing access to the cx register. I am not into hacking gcc's register allocator 8) -- Amancio Hasty [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
