Greetings!

[ Got this off the maxima list, did you cc me directly? ]

David Scherfgen via Maxima-discuss
<maxima-disc...@lists.sourceforge.net> writes:

> Hmm, not sure if this is helpful:
> 
> Reading symbols from ../unixport/raw_pre_gcl...
> (gdb) set env GCL_LSPSYSDIR .././unixport/
> (gdb) set env GCL_MEM_BOUND 29
> (gdb) b Icall_gen_error_handler_noreturn
> Breakpoint 1 at 0xb5b50: file o/error.c, line 165.
> (gdb) r
> Starting program: /home/gcl-builder/gcl/gcl/unixport/raw_pre_gcl
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
> GCL (GNU Common Lisp)  April 1994  516 pages
> Building symbol table for /home/gcl-builder/gcl/gcl/unixport/raw_pre_gcl ..
> loading .././unixport/../lsp/gcl_export.lsp
> loading .././unixport/../lsp/gcl_defmacro.lsp
> loading .././unixport/../lsp/gcl_evalmacros.lsp
> loading .././unixport/../lsp/gcl_top.lsp
> loading .././unixport/../lsp/gcl_autoload.lsp
> 
> >bt
> 
> Breakpoint 1, Icall_gen_error_handler_noreturn (ci=0x32d278 <Cnil_body>,
> cs=0xc23900, en=0x12023c8, es=0xc23900, n=n@entry=2) at o/error.c:165
> 165     Icall_gen_error_handler_noreturn(object ci,object cs,object
> en,object es,ufixnum n,...) {
> 
> Am Fr., 25. Apr. 2025 um 23:54 Uhr schrieb Camm Maguire <
> ca...@ma...>:


Yes, this is a perfectly normal run!  This should build fine.  Perhaps
there is an issue with your C compiler and 'full' optimization?  Can you
complete the build with

./configure CFLAGS="-g -O2" ... (whatever else you need)

The 'bt' was assuming you would hit the error handler, which you did
not.  It is a gdb command, not a gcl one.

It is also relevant if the hand compile command I gave you produces the
.c files or not, so if you could let me know....  As the other poster
has indicated, there may be some instability if your build tree
traverses links in uncommon places, which should be tracked down
eventually.  Your symptoms seem different, however, as in that case
problematic paths with double slashes were produced, triggering an
intelligible error.

GCL has installed the GNU build system, which means it can compile in a
fresh build directory and leave the source directory untouched if
desired:

mkdir build
cd build
../configure .... && make

There is only one master Makefile.am in the top directory -- no
recursive make.  All paths are indicated explicitly in each target.  As
your build directory and source directory are the same, this is likely
not relevant, but the system will place compiled products in the build
directory, and read the sources from the source directory, so you may
want to `find -name gcl_s.c` to see if the file is misplaced somehow.

I understand the random nature of your failures does not allow for the
.c file detection just yet, but eventually....

Take care,


David Scherfgen via Maxima-discuss
<maxima-disc...@lists.sourceforge.net> writes:

> I sent the make.log showing the gcl_typep.lsp crash yesterday, here it is 
> again: https://pastebin.com/EqkjDwaF
> This is the crash where I used GDB as you requested and it showed an endless 
> recursion:
>
> #0  0x00000000000a7e69 in maybe_reallocate_page ()
> #1  0x00000000000afbc5 in alloc_object ()
> #2  0x0000000000098ba8 in alloc_simple_string ()
> #3  0x0000000000098f21 in make_simple_string ()
> #4  0x00000000000afc55 in alloc_object ()
> #5  0x0000000000098ba8 in alloc_simple_string ()
> #6  0x0000000000098f21 in make_simple_string ()
> #7  0x00000000000afc55 in alloc_object ()
> #8  0x0000000000098ba8 in alloc_simple_string ()
> #9  0x0000000000098f21 in make_simple_string ()
> #10 0x00000000000afc55 in alloc_object ()
> [repeating for a very long time]
>
> Unfortunately, I can't do unixport/saved_gcl3 right now, since I don't manage 
> to get far enough with the build any more ... Maybe it's just luck and I have 
> to keep
> trying.
>
> Am Fr., 25. Apr. 2025 um 23:11 Uhr schrieb Camm Maguire 
> <c...@maguirefamily.org>:
>
>  Greetings!
>
>  Clearly something non-reproducible is going on.
>
>  Please do
>
>  unixport/saved_gcl3
>  >(compile-file "lsp/gcl_s.lsp")
>  >(si::bye)
>  ls -lrt lsp/gcl_s.*
>
>  If you do not see .lsp, .o, .c, .h, and .data files, then
>
>  unixport/saved_gcl3
>  >compiler::*default-c-file*
>  >compiler::*default-h-file*
>  >compiler::*default-data-file*
>  >compiler::*default-system-p*
>
>  I don't know what to say about the crash at gcl_typep.lsp, as I don't
>  think I've seen a log showing this (yet).
>
>  Take care,
>
>  David Scherfgen via Maxima-discuss
>  <maxima-disc...@lists.sourceforge.net> writes:
>
>  > Yes indeed, unixport/saved_ansi_gcl seems to be working!
>  >
>  > $ ls -lrt gcl/gcl_s.*
>  > -rw-rw-r-- 1 gcl-builder gcl-builder 61813 Apr 25 08:32 gcl/gcl_s.o
>  > -rw-rw-r-- 1 gcl-builder gcl-builder 43685 Apr 25 08:32 gcl/gcl_s.data
>  >
>  > $ ls -lrt gcl3/gcl_s.*
>  > -rw-rw-r-- 1 gcl-builder gcl-builder 43685 Apr 25 08:32 gcl3/gcl_s.data
>  > -rw-rw-r-- 1 gcl-builder gcl-builder 61813 Apr 25 08:32 gcl3/gcl_s.o
>  >
>  > Regarding the missing file, a hardware issue seems unlikely, as I'm using 
> ZFS on a 2x NVMe mirror.
>  >
>  > Also strange: I just tried to build again with exactly the same setup as 
> before, and this time it crashed while compiling lap/gcl_typep.lsp, just like
>  yesterday.
>  >
>  > Another attempt, after rebooting, resulted in the same problem.
>  >
>  > Camm Maguire <c...@maguirefamily.org> schrieb am Fr., 25. Apr. 2025, 22:01:
>  >
>  >  Greetings!  Your build basically finished but for the final profiling
>  >  images.  You likely have a working unixport/saved_ansi_gcl.
>  >
>  >  The error seems to point to some os or hardware failure
>  >  writing/retaining a file.  Please send `ls -lrt gcl/gcl_s.*` and `ls
>  >  -lrt gcl3/gcl_s.*`.  The .o file is there, but you seem to be missing
>  >  the .c file.  And your log shows 'cp gcl3/gcl_s.* gcl' succeeding.
>  >
>  >  Take care,
>  >
>  >  David Scherfgen via Maxima-discuss
>  >  <maxima-disc...@lists.sourceforge.net> writes:
>  >
>  >  > Oh, now it fails at a different point.
>  >  > config.log: https://pastebin.com/vvLeHVaK
>  >  > make.log: https://pastebin.eldshort.de/bQyxGMAP_4YlYA1-hhzyK (exceeds 
> pastebin.com's size limit)
>  >  > The error is now: "cc1: fatal error: gcl/gcl_s.c: No such file or 
> directory"
>  >  >
>  >  > Camm Maguire <c...@maguirefamily.org> schrieb am Fr., 25. Apr. 2025, 
> 20:11:
>  >  >
>  >  >  Greetings!  I believe your previous attempt failed earlier than this.
>  >  >  Can you please refresh the pastebins?
>  >  >
>  >  >  Take care,
>  >  >
>  >  >  David Scherfgen via Maxima-discuss
>  >  >  <maxima-disc...@lists.sourceforge.net> writes:
>  >  >
>  >  >  > Unfortunately, configure --enable-min-pagewidth=21 didn't help. It 
> still crashes the same way while compiling gcl_typep.lsp. Of course I made 
> sure
>  to
>  >  >  clean up
>  >  >  > everything before configuring and making.
>  >  >  >
>  >  >  > Camm Maguire <c...@maguirefamily.org> schrieb am Fr., 25. Apr. 2025, 
> 04:38:
>  >  >  >
>  >  >  >  Greetings!  OK, I've just tested this, and thankfully you can just 
> use
>  >  >  >  2.7.1 with a configure option:
>  >  >  >
>  >  >  >  ./configure --enable-min-pagewidth=21
>  >  >  >
>  >  >  >  will override the system settings and set your pagesize to (1<<21)
>  >  >  >  bytes.   I thought this would cause the madvise call to fail, but it
>  >  >  >  doesn't.
>  >  >  >
>  >  >  >  Option needs renaming, but that is for later...
>  >  >  >
>  >  >  >  Please let me know if problems persist.
>  >  >  >
>  >  >  >  Take care,
>  >  >  >  -- 
>  >  >  >  Camm Maguire                                        
> c...@maguirefamily.org
>  >  >  >  
> ==========================================================================
>  >  >  >  "The earth is but one country, and mankind its citizens."  --  
> Baha'u'llah
>  >  >  >
>  >  >  > _______________________________________________
>  >  >  > Maxima-discuss mailing list
>  >  >  > maxima-disc...@lists.sourceforge.net
>  >  >  > https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>  >  >  >
>  >  >
>  >  >  -- 
>  >  >  Camm Maguire                                        
> c...@maguirefamily.org
>  >  >  
> ==========================================================================
>  >  >  "The earth is but one country, and mankind its citizens."  --  
> Baha'u'llah
>  >  >
>  >  > _______________________________________________
>  >  > Maxima-discuss mailing list
>  >  > maxima-disc...@lists.sourceforge.net
>  >  > https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>  >  >
>  >
>  >  -- 
>  >  Camm Maguire                                        c...@maguirefamily.org
>  >  ==========================================================================
>  >  "The earth is but one country, and mankind its citizens."  --  Baha'u'llah
>  >
>  > _______________________________________________
>  > Maxima-discuss mailing list
>  > maxima-disc...@lists.sourceforge.net
>  > https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>  >
>
>  -- 
>  Camm Maguire                                        c...@maguirefamily.org
>  ==========================================================================
>  "The earth is but one country, and mankind its citizens."  --  Baha'u'llah
>
> _______________________________________________
> Maxima-discuss mailing list
> maxima-disc...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>

-- 
Camm Maguire                                        c...@maguirefamily.org
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah

Reply via email to