[Bug c/95652] New: GCC 8.3.1 generates syntactically incorrect assembly code

[teo.samarzija at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95652

Bug ID: 95652
   Summary: GCC 8.3.1 generates syntactically incorrect assembly
code
   Product: gcc
   Version: 8.3.1
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: c
  Assignee: unassigned at gcc dot gnu.org
  Reporter: teo.samarzija at gmail dot com
  Target Milestone: ---

Hey, guys!

So, I think I've found a bug in GCC 8.3.1 that occurs on Linux. Here is a
simple test case:

int printf(const char*,...);

float eax;

int main() {
eax=7;
printf("%lf\n",eax);
}

When compiled with "gcc -masm=intel -o test test.c", I get the error messages:

/tmp/ccGiemfM.s: Assembler messages:
/tmp/ccGiemfM.s:20: Error: invalid use of register
/tmp/ccGiemfM.s:21: Error: invalid use of register

It appears as though GCC 8.3.1 generates syntactically incorrect assembly code
for that.
Possibly related: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52554

[Bug c/95652] GCC 8.3.1 generates syntactically incorrect assembly code

[teo.samarzija at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95652

--- Comment #1 from Teo Samarzija  ---
Possibly related, "as --version" prints:

GNU assembler (GNU Binutils) 2.33.1
Copyright (C) 2019 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public License version 3 or later.
This program has absolutely no warranty.
This assembler was configured for a target of `x86_64-pc-linux-gnu'.

[Bug target/95652] GCC 8.3.1 generates syntactically incorrect assembly code with -masm=intel

[teo.samarzija at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95652

--- Comment #3 from Teo Samarzija  ---
(In reply to Andrew Pinski from comment #2)
> This looks closer to a dup of bug 87986.
> 
> Basically -masm=intel is not always working.

When does it occur? Maybe we can determine when it can occur and warn about it?
As far I can tell, it occurs when you use a reserved word in assembly as the
name of the global variable, and only under Linux (not under Windows, since the
names of global variables on Windows get prepended '_').
Why exactly does GCC use such a different convention for Windows and Linux?
Perhaps you can change that.
Besides, how does CLANG compile that same code fine, also under Linux and with
"-masm=intel"? Maybe you can copy the way CLANG does that.

[Bug target/95652] GCC 8.3.1 generates syntactically incorrect assembly code with -masm=intel

[teo.samarzija at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95652

--- Comment #5 from Teo Samarzija  ---
(In reply to Uroš Bizjak from comment #4)
> (In reply to Teo Samarzija from comment #3)
> > Besides, how does CLANG compile that same code fine, also under Linux and
> > with "-masm=intel"? Maybe you can copy the way CLANG does that.
> 
> Clang outputs:
> 
>   mov dword ptr [rip + eax], 1088421888
> 
> which gld does't understand:
> 
> pr95652.s: Assembler messages:
> pr95652.s:17: Error: `dword ptr [rip+eax]' is not a valid base/index
> expression

Which version of CLANG? CLANG 9.0 compiles the code in the opening post
correctly (under Linux and using Intel Syntax).

[Bug c/95800] New: DJGPP 9.3.1 doesn't parse C files correctly

[teo.samarzija at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95800

Bug ID: 95800
   Summary: DJGPP 9.3.1 doesn't parse C files correctly
   Product: gcc
   Version: 9.3.1
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: c
  Assignee: unassigned at gcc dot gnu.org
  Reporter: teo.samarzija at gmail dot com
  Target Milestone: ---

So, the C library I am using to cross-compile programs from Linux to DOS
apparently doesn't support the "log2" function, and I tried to polyfill it:

#include 

double log2(double x)
{
return log(x)/log(2);
}

int main() {
return 0;
}

However, when I do that, I get the following error, and the linking doesn't
even begin:
test.c:3:8: error: expected ')' before '/' token
3 | double log2(double x)
  |^~~~

Here is what "djgpp-gcc -v" prints:
Using built-in specs.
COLLECT_GCC=djgpp-gcc
COLLECT_LTO_WRAPPER=/home/teo.samarzija/djgpp-9.3.0/libexec/gcc/djgpp/9.3.0/lto-wrapper
Target: djgpp
Configured with: ../gcc-9.3.0/configure --target=djgpp
--prefix=/home/teo.samarzija/djgpp-9.3.0
--enable-languages=c,c++,objc,ada,fortran,go
Thread model: single
gcc version 9.3.0 (GCC)

[Bug c/95800] DJGPP 9.3.1 doesn't parse C files correctly

[teo.samarzija at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95800

--- Comment #2 from Teo Samarzija  ---
(In reply to Jakub Jelinek from comment #1)
> Perhaps log2 is in DJGPP a macro (which the C standard allows)?
> In that case, you either need to #undef that macro, or
> use double (log2)(double x) { ... }

I am sorry, apparently it is defined as a macro. I don't know why I am getting
linker errors telling me "log2" is undefined.