Re: print_npx_status_word

Re: print_npx_status_word

[Baars, M.J.]

[-- Attachment #1: Type: text/plain, Size: 294 bytes --]

On 04/15/2011 06:40 PM, Tom Tromey wrote:
> I don't know.
>
> It is definitely there:
>
> http://sourceware.org/ml/gdb/2011-04/msg00063.html
>
> Tom

Hi Tom,

I believe this is the format you requested?

Do you also have the FSF copyright forms that I will need to fill out?

Regards,
Michael


[-- Attachment #2: patches-gdb-7.2.tar.gz --]
[-- Type: application/x-gzip, Size: 2521 bytes --]

Re: print_npx_status_word

[Mark Kettenis]

> Date: Sat, 16 Apr 2011 08:08:48 +0200
> From: "Baars, M.J." <mjbaars1977.gdb@gmail.com>
> 
> Hi Tom,
> 
> I believe this is the format you requested?

No!  This is a .tar.gz.  Proper diffs don't come as a .tar.gz.

> --------------020309050206070002040202
> Content-Type: application/x-gzip;
>  name="patches-gdb-7.2.tar.gz"
> Content-Transfer-Encoding: base64
> Content-Disposition: attachment;
>  filename="patches-gdb-7.2.tar.gz"

Re: print_npx_status_word

[Baars, M.J.]

[-- Attachment #1: Type: text/plain, Size: 554 bytes --]

On 04/16/2011 10:24 AM, Mark Kettenis wrote:
>> Date: Sat, 16 Apr 2011 08:08:48 +0200
>> From: "Baars, M.J."<mjbaars1977.gdb@gmail.com>
>>
>> Hi Tom,
>>
>> I believe this is the format you requested?
> No!  This is a .tar.gz.  Proper diffs don't come as a .tar.gz.
Oops... did I remove one of your entries in the ChangeLog? :)
>> --------------020309050206070002040202
>> Content-Type: application/x-gzip;
>>   name="patches-gdb-7.2.tar.gz"
>> Content-Transfer-Encoding: base64
>> Content-Disposition: attachment;
>>   filename="patches-gdb-7.2.tar.gz"


[-- Attachment #2: ChangeLog.diff --]
[-- Type: text/x-patch, Size: 382 bytes --]

--- ./gdb-7.2/gdb/ChangeLog	2010-09-03 01:37:25.000000000 +0200
+++ ./gdb-7.2.2/gdb/ChangeLog	2011-04-14 16:14:53.623948000 +0200
@@ -1,3 +1,7 @@
+2011-04-14	Ir. M.J. Baars <mjbaars1977.gdb@gmail.com>
+
+	* i387-tdep.c (print_npx_status_word) - new function
+
 2010-09-02  Joel Brobecker  <brobecker@adacore.com>
 
 	* NEWS: Replace "Changes since GDB 7.1" by "Changes in GDB 7.2".

[-- Attachment #3: ChangeLog-1999.diff --]
[-- Type: text/x-patch, Size: 704 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-1999	2001-01-12 04:44:31.000000000 +0100
+++ ./gdb-7.2.2/gdb/ChangeLog-1999	2011-04-14 16:11:31.503948001 +0200
@@ -581,9 +581,9 @@ Wed Dec  8 15:29:48 1999  Andrew Cagney
 	Patch from Mark Kettenis  <kettenis@gnu.org>:
 
 	* config/i386/tm-i386.h (FLOAT_INFO): New define.
-	* i387-tdep.c (print_i387_value, print_i387_ext,
-	print_i387_status_word, print_i387_control_word, i387_float_info):
-	New functions, used to implement generic `info float' command.
+	
+	* i387-tdep.c (print_i387_value, print_i387_ext,print_i387_control_word, i387_float_info):
+	  New functions, used to implement generic `info float' command.
 
 1999-12-06  Christopher Faylor  <cgf@cygnus.com>
 

[-- Attachment #4: ChangeLog-2001.diff --]
[-- Type: text/x-patch, Size: 557 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-2001	2002-01-06 05:37:53.000000000 +0100
+++ ./gdb-7.2.2/gdb/ChangeLog-2001	2011-04-14 15:55:22.631948002 +0200
@@ -8076,11 +8076,6 @@ Wed Apr  4 21:48:42 2001  Christopher Fa
 
 	* MAINTAINERS: Add myself to write after approval list.
 
-2001-03-09  Mark Kettenis  <kettenis@gnu.org>
-
-	* i387-tdep.c (print_i387_status_word): Fix printing of Stack
-	fault flag.  It is bit 6 of the x87 FPU status word, not bit 7.
-
 2001-03-08  Andrew Cagney  <ac131313@redhat.com>
 
 	* configure.host (m88*-harris-cxux*): Mark as obsolete.

[-- Attachment #5: ChangeLog-2002.diff --]
[-- Type: text/x-patch, Size: 1081 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-2002	2004-04-24 00:15:17.000000000 +0200
+++ ./gdb-7.2.2/gdb/ChangeLog-2002	2011-04-14 16:13:29.187948001 +0200
@@ -5396,12 +5396,13 @@ Fri Sep 13 07:42:09 2002  Andrew Cagney
 	* config/i386/tm-i386.h: Don't include "regcache.h".
 
 	* i387-tdep.h (i387_print_float_info): New prototype.
-	* i387-tdep.c (print_i387_value, print_i387_ext,
-	print_i387_status_word, print_i387_control_word): Add `struct
-	ui_file *' argument and use it for output.
-	(i387_print_float_info): Renamed from i387_float_info.  Add
-	`struct gdbarch *' and `struct ui_file *' arguments and use the
-	latter for output.
+	* i387-tdep.c (print_i387_value, print_i387_ext, print_i387_control_word):
+	  Add `struct ui_file *' argument and use it for output.
+
+	* i387-tdep.c (i387_print_float_info):
+	  Renamed from i387_float_info.  Add `struct gdbarch *' and `struct ui_file *' arguments
+	  and use the latter for output.
+
 	* i386-tdep.c: Include "i387-tdep.h".
 	(i386_gdbarch_init): Set print_float_info.
 	* config/i386/tm-i386.h (i387_float_info): Remove prototype.

[-- Attachment #6: ChangeLog-2004.diff --]
[-- Type: text/x-patch, Size: 679 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-2004	2006-10-24 22:23:56.000000000 +0200
+++ ./gdb-7.2.2/gdb/ChangeLog-2004	2011-04-14 16:13:52.067948001 +0200
@@ -3468,9 +3468,10 @@
 	* ui-out.c (ui_out_field_core_addr): Ditto.
 	* breakpoint.c (breakpoint_adjustment_warning): Ditto.
 	* exec.c (print_section_info): Ditto.
-	* i387-tdep.c (print_i387_status_word): Ditto.
-	(print_i387_control_word): Ditto.
-	(i387_print_float_info): Ditto.
+
+	* i387-tdep.c (print_i387_control_word): Ditto.
+	* i387-tdep.c (i387_print_float_info): Ditto.
+
 	* maint.c (maint_print_section_info): Ditto.
 	* solib.c (info_sharedlibrary_command): Ditto.
 	* somsolib.c (som_sharedlibrary_info_command): Ditto.

[-- Attachment #7: i387-tdep.c.diff --]
[-- Type: text/x-patch, Size: 4261 bytes --]

--- ./gdb-7.2/gdb/i387-tdep.c	2010-05-05 21:30:36.000000000 +0200
+++ ./gdb-7.2.2/gdb/i387-tdep.c	2011-04-14 16:23:54.647948001 +0200
@@ -36,6 +36,52 @@
 #include "i387-tdep.h"
 #include "i386-xstate.h"
 
+struct
+{
+	char *tag;
+	char *description;
+}
+
+static const npx_exception_flags[6] =
+
+{
+	{"IE", "invalid operation"},
+	{"DE", "denormalized operand"},
+	{"ZE", "zero divide"},
+	{"OE", "overflow"},
+	{"UE", "underflow"},
+	{"PE", "precision"}
+};
+
+// print the numeric coprocessor extension (npx) status word
+void print_npx_status_word(uint16_t npx_status_word, struct ui_file *file)
+{
+	fprintf_filtered(file, "status word              : 0x%04X\n", npx_status_word);
+
+	fprintf_filtered(file, "  exception flags        : ");
+
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0005) & 0x0001) ? npx_exception_flags[5].tag : "  ");	// precision
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0004) & 0x0001) ? npx_exception_flags[4].tag : "  ");	// underflow
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0003) & 0x0001) ? npx_exception_flags[3].tag : "  ");	// overflow
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0002) & 0x0001) ? npx_exception_flags[2].tag : "  ");	// zero divide
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0001) & 0x0001) ? npx_exception_flags[1].tag : "  ");	// denormalized operand
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0000) & 0x0001) ? npx_exception_flags[0].tag : "  ");	// invalid operation
+	fprintf_filtered(file, "\n");
+
+	fprintf_filtered(file, "  stack fault            : %u\n", ((npx_status_word >> 0x0006) & 0x0001));
+	fprintf_filtered(file, "  error summary status   : %u\n", ((npx_status_word >> 0x0007) & 0x0001));
+	fprintf_filtered(file, "  busy                   : %u\n", ((npx_status_word >> 0x000F) & 0x0001));
+	fprintf_filtered(file, "  top of stack pointer   : %u\n", ((npx_status_word >> 0x000B) & 0x0007));
+
+	fprintf_filtered(file, "  condition code         : ");
+
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x000E) & 0x0001) ? "C3" : "  ");
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x000A) & 0x0001) ? "C2" : "  ");
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0009) & 0x0001) ? "C1" : "  ");
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0008) & 0x0001) ? "C0" : "  ");
+	fprintf_filtered(file, "\n");
+}
+
 /* Print the floating point number specified by RAW.  */
 
 static void
@@ -114,36 +160,6 @@ print_i387_ext (struct gdbarch *gdbarch,
     fputs_filtered (" Unsupported", file);
 }
 
-/* Print the status word STATUS.  */
-
-static void
-print_i387_status_word (unsigned int status, struct ui_file *file)
-{
-  fprintf_filtered (file, "Status Word:         %s",
-		    hex_string_custom (status, 4));
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0001) ? "IE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0002) ? "DE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0004) ? "ZE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0008) ? "OE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0010) ? "UE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0020) ? "PE" : "  ");
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0080) ? "ES" : "  ");
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0040) ? "SF" : "  ");
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0100) ? "C0" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0200) ? "C1" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0400) ? "C2" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x4000) ? "C3" : "  ");
-
-  fputs_filtered ("\n", file);
-
-  fprintf_filtered (file,
-		    "                       TOP: %d\n", ((status >> 11) & 7));
-}
-
 /* Print the control word CONTROL.  */
 
 static void
@@ -268,7 +284,9 @@ i387_print_float_info (struct gdbarch *g
 
   fputs_filtered ("\n", file);
 
-  print_i387_status_word (fstat, file);
+	print_npx_status_word(fstat, file);
+	fprintf_filtered(file, "\n");
+
   print_i387_control_word (fctrl, file);
   fprintf_filtered (file, "Tag Word:            %s\n",
 		    hex_string_custom (ftag, 4));

Re: print_npx_status_word

[Daniel Jacobowitz]

On Sat, Apr 16, 2011 at 03:46:09PM +0200, Baars, M.J. wrote:
> On 04/16/2011 10:24 AM, Mark Kettenis wrote:
> >>Date: Sat, 16 Apr 2011 08:08:48 +0200
> >>From: "Baars, M.J."<mjbaars1977.gdb@gmail.com>
> >>
> >>Hi Tom,
> >>
> >>I believe this is the format you requested?
> >No!  This is a .tar.gz.  Proper diffs don't come as a .tar.gz.
> Oops... did I remove one of your entries in the ChangeLog? :)

If you want a patch to be committed, eventually you're going to have
to say why.  What's the goal of the patch?

Also, why on earth are you messing about with old changelog entries?

-- 
Daniel Jacobowitz
Mentor Graphics

Re: print_npx_status_word

[Baars, M.J.]

On 04/16/2011 04:29 PM, Daniel Jacobowitz wrote:
> On Sat, Apr 16, 2011 at 03:46:09PM +0200, Baars, M.J. wrote:
>> On 04/16/2011 10:24 AM, Mark Kettenis wrote:
>>>> Date: Sat, 16 Apr 2011 08:08:48 +0200
>>>> From: "Baars, M.J."<mjbaars1977.gdb@gmail.com>
>>>>
>>>> Hi Tom,
>>>>
>>>> I believe this is the format you requested?
>>> No!  This is a .tar.gz.  Proper diffs don't come as a .tar.gz.
>> Oops... did I remove one of your entries in the ChangeLog? :)
> If you want a patch to be committed, eventually you're going to have
> to say why.  What's the goal of the patch?
When you look at the code in question more closely, you will see that it 
will present the content of the status word in more detail than the 
original function,
> Also, why on earth are you messing about with old changelog entries?
>
The old function can thus be seen as obsolete, and should be removed 
from the old log entries.

Regards,
Michael.

Re: print_npx_status_word

[Mark Kettenis]

> Date: Sat, 16 Apr 2011 18:21:16 +0200
> From: "Baars, M.J." <mjbaars1977.gdb@gmail.com>
> 
> On 04/16/2011 04:29 PM, Daniel Jacobowitz wrote:
> > On Sat, Apr 16, 2011 at 03:46:09PM +0200, Baars, M.J. wrote:
> >> On 04/16/2011 10:24 AM, Mark Kettenis wrote:
> >>>> Date: Sat, 16 Apr 2011 08:08:48 +0200
> >>>> From: "Baars, M.J."<mjbaars1977.gdb@gmail.com>
> >>>>
> >>>> Hi Tom,
> >>>>
> >>>> I believe this is the format you requested?
> >>> No!  This is a .tar.gz.  Proper diffs don't come as a .tar.gz.
> >> Oops... did I remove one of your entries in the ChangeLog? :)
> > If you want a patch to be committed, eventually you're going to have
> > to say why.  What's the goal of the patch?
> When you look at the code in question more closely, you will see that it 
> will present the content of the status word in more detail than the 
> original function,

Which isn't necessarily a good thing.

> > Also, why on earth are you messing about with old changelog entries?
> >
> The old function can thus be seen as obsolete, and should be removed 
> from the old log entries.

Seems you totally fail to understand what ChangeLogs are for.

Re: print_npx_status_word

[Eli Zaretskii]

> Date: Sat, 16 Apr 2011 18:21:16 +0200
> From: "Baars, M.J." <mjbaars1977.gdb@gmail.com>
> 
> > Also, why on earth are you messing about with old changelog entries?
> >
> The old function can thus be seen as obsolete, and should be removed 
> from the old log entries.

ChangeLog files are the record of GDB history.  We don't rewrite
history when we replace functions by better ones.  So please leave the
old log entries alone, and just add new ones.

Re: print_npx_status_word

[Baars, M.J.]

On 04/16/2011 06:30 PM, Mark Kettenis wrote:
>> Date: Sat, 16 Apr 2011 18:21:16 +0200
>> From: "Baars, M.J."<mjbaars1977.gdb@gmail.com>
>>
>> On 04/16/2011 04:29 PM, Daniel Jacobowitz wrote:
>>> On Sat, Apr 16, 2011 at 03:46:09PM +0200, Baars, M.J. wrote:
>>>> On 04/16/2011 10:24 AM, Mark Kettenis wrote:
>>>>>> Date: Sat, 16 Apr 2011 08:08:48 +0200
>>>>>> From: "Baars, M.J."<mjbaars1977.gdb@gmail.com>
>>>>>>
>>>>>> Hi Tom,
>>>>>>
>>>>>> I believe this is the format you requested?
>>>>> No!  This is a .tar.gz.  Proper diffs don't come as a .tar.gz.
>>>> Oops... did I remove one of your entries in the ChangeLog? :)
>>> If you want a patch to be committed, eventually you're going to have
>>> to say why.  What's the goal of the patch?
>> When you look at the code in question more closely, you will see that it
>> will present the content of the status word in more detail than the
>> original function,
> Which isn't necessarily a good thing.

I'll show you the difference in output after the 'info float' command.

original output:

Status Word:         0x0000
                        TOP: 0

new and improved output:

status word              : 0x0000
   exception flags        :
   stack fault            : 0
   error summary status   : 0
   busy                   : 0
   top of stack pointer   : 0
   condition code         :

This will give you a much better overview of what is actually happing in 
the numeric core, which is eventually what every programmer wants is 
what I was always taught.

>>> Also, why on earth are you messing about with old changelog entries?
>>>
>> The old function can thus be seen as obsolete, and should be removed
>> from the old log entries.
> Seems you totally fail to understand what ChangeLogs are for.

Would you rather have obsolete function names flying all over the place 
then? I don't understand, eventually this will get you into trouble.

Regards,
Michael

Re: print_npx_status_word

[Mark Kettenis]

> Date: Sat, 16 Apr 2011 18:46:54 +0200
> From: "Baars, M.J." <mjbaars1977.gdb@gmail.com>
>
> >> When you look at the code in question more closely, you will see that it
> >> will present the content of the status word in more detail than the
> >> original function,
> > Which isn't necessarily a good thing.
> 
> I'll show you the difference in output after the 'info float' command.
> 
> original output:
> 
> Status Word:         0x0000
>                         TOP: 0
> 
> new and improved output:
> 
> status word              : 0x0000
>    exception flags        :
>    stack fault            : 0
>    error summary status   : 0
>    busy                   : 0
>    top of stack pointer   : 0
>    condition code         :
> 
> This will give you a much better overview of what is actually happing in 
> the numeric core, which is eventually what every programmer wants is 
> what I was always taught.

But it will add a lot of extra lines to the output of "info float"
such that it doesn't fit on a standard 24-line screen anymore.  The
current way of printing things is much more concise.  And your way of
printing things doesn't really add more information; it just uses more
words.  Admittedly to interpret the printed codes you probably need to
have the IA-32 Architecture Software Developer's manual at hand.  But
to be honest, if you're looking at the FPU at this detailed level you
probably already are reading that manual.

Re: print_npx_status_word

[Baars, M.J.]

On 04/16/2011 07:09 PM, Mark Kettenis wrote:
>> Date: Sat, 16 Apr 2011 18:46:54 +0200
>> From: "Baars, M.J."<mjbaars1977.gdb@gmail.com>
>>
>>>> When you look at the code in question more closely, you will see that it
>>>> will present the content of the status word in more detail than the
>>>> original function,
>>> Which isn't necessarily a good thing.
>> I'll show you the difference in output after the 'info float' command.
>>
>> original output:
>>
>> Status Word:         0x0000
>>                          TOP: 0
>>
>> new and improved output:
>>
>> status word              : 0x0000
>>     exception flags        :
>>     stack fault            : 0
>>     error summary status   : 0
>>     busy                   : 0
>>     top of stack pointer   : 0
>>     condition code         :
>>
>> This will give you a much better overview of what is actually happing in
>> the numeric core, which is eventually what every programmer wants is
>> what I was always taught.
> But it will add a lot of extra lines to the output of "info float"
> such that it doesn't fit on a standard 24-line screen anymore.  The
I suppose you mean 50 lines? That should fit! Are we talking about the 
same standard screen?
> current way of printing things is much more concise.  And your way of
> printing things doesn't really add more information; it just uses more
> words.  Admittedly to interpret the printed codes you probably need to
> have the IA-32 Architecture Software Developer's manual at hand.  But
> to be honest, if you're looking at the FPU at this detailed level you
> probably already are reading that manual.
The exception flags and exception codes are indeed not explained in such 
detail on the screen as they are in the manual. I could implement an 
extra help function to that end later on if you like.

Regards,
Michael

Re: print_npx_status_word

[Marek Polacek]

On 04/16/2011 07:55 PM, Baars, M.J. wrote:
> I could implement an extra help function to that end later on if you like.

Please don't do this.  This would be only bloat.

Re: print_npx_status_word

[Tom Tromey]

>>>>> "Mark" == Mark Kettenis <mark.kettenis@xs4all.nl> writes:

Mark> But it will add a lot of extra lines to the output of "info float"
Mark> such that it doesn't fit on a standard 24-line screen anymore.  The
Mark> current way of printing things is much more concise.  And your way of
Mark> printing things doesn't really add more information; it just uses more
Mark> words.

Something like `info float verbose' or `info float/v' seems ok to me.

Tom

Re: print_npx_status_word

[Tom Tromey]

>>>>> "Michael" == Baars, M J <mjbaars1977.gdb@gmail.com> writes:

Michael> Do you also have the FSF copyright forms that I will need to
Michael> fill out?

I will send them separately.

Tom

Re: print_npx_status_word

[Tom Tromey]

>>>>> "Michael" == Baars, M J <mjbaars1977.gdb@gmail.com> writes:

Michael> Oops... did I remove one of your entries in the ChangeLog? :)

As others have noted, ChangeLog is "prepend only".
We generally don't edit old entries.

Michael> +struct
Michael> +{
Michael> +	char *tag;
Michael> +	char *description;
Michael> +}

Wrong indentation here, should be 2 spaces.
Fields should have descriptive comments.
I think the struct should have a tag and be terminated here.
Also the fields should be "const".

Michael> +static const npx_exception_flags[6] =
Michael> +
Michael> +{
Michael> +	{"IE", "invalid operation"},

Wrong indentation.  This is a problem in the whole patch.

Michael> +// print the numeric coprocessor extension (npx) status word

No `//' comments.
Comments should be full sentences, see GNU standards.

Michael> +void print_npx_status_word(uint16_t npx_status_word, struct ui_file *file)

Newline after "void".
Space before "(" -- a problem in a few spots.
Again, GNU standards.

Tom

Re: print_npx_status_word

[Baars, M.J.]

[-- Attachment #1: Type: text/plain, Size: 1803 bytes --]

On 04/18/2011 07:03 PM, Tom Tromey wrote:
>>>>>> "Michael" == Baars, M J<mjbaars1977.gdb@gmail.com>  writes:
Hi Tom,

Thank you for your reply, for your convenience I attached a slightly 
different version already.
> Michael>  Oops... did I remove one of your entries in the ChangeLog? :)
>
> As others have noted, ChangeLog is "prepend only".
> We generally don't edit old entries.
As I stated before, it would be unwise though to keep unmaintained 
function names in the log entries. You can apply the changes optionally, 
but I strongly advise against retaining the original log entries.
> Michael>  +struct
> Michael>  +{
> Michael>  +	char *tag;
> Michael>  +	char *description;
> Michael>  +}
>
> Wrong indentation here, should be 2 spaces.
> Fields should have descriptive comments.
> I think the struct should have a tag and be terminated here.
> Also the fields should be "const".
Here you are completely right, I added the 'const' operators. If you 
don't like the indentation, I suggest you instead modify you 
tab-settings locally. Personally I'm used programming with a little more 
space than.
> Michael>  +static const npx_exception_flags[6] =
> Michael>  +
> Michael>  +{
> Michael>  +	{"IE", "invalid operation"},
>
> Wrong indentation.  This is a problem in the whole patch.
>
> Michael>  +// print the numeric coprocessor extension (npx) status word
>
> No `//' comments.
I have taken this into account, does that also hold for inline comments?
> Comments should be full sentences, see GNU standards.
>
> Michael>  +void print_npx_status_word(uint16_t npx_status_word, struct ui_file *file)
>
> Newline after "void".
> Space before "(" -- a problem in a few spots.
> Again, GNU standards.
Hope you'll make an exception here, I like the code better this way :)
> Tom
Regards,
Michael.

[-- Attachment #2: ChangeLog.diff --]
[-- Type: text/x-patch, Size: 498 bytes --]

--- ./gdb-7.2/gdb/ChangeLog	2010-09-03 01:37:25.000000000 +0200
+++ ./gdb-7.2.2/gdb/ChangeLog	2011-04-16 22:56:35.359736001 +0200
@@ -1,3 +1,11 @@
+2011-04-16	Ir. M.J. Baars <mjbaars1977.gdb@gmail.com>
+
+	* i387-tdep.c (print_npx_control_word) - new function
+
+2011-04-14	Ir. M.J. Baars <mjbaars1977.gdb@gmail.com>
+
+	* i387-tdep.c (print_npx_status_word) - new function
+
 2010-09-02  Joel Brobecker  <brobecker@adacore.com>
 
 	* NEWS: Replace "Changes since GDB 7.1" by "Changes in GDB 7.2".

[-- Attachment #3: ChangeLog-1999.diff --]
[-- Type: text/x-patch, Size: 678 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-1999	2001-01-12 04:44:31.000000000 +0100
+++ ./gdb-7.2.2/gdb/ChangeLog-1999	2011-04-16 22:53:25.543736000 +0200
@@ -581,9 +581,9 @@ Wed Dec  8 15:29:48 1999  Andrew Cagney
 	Patch from Mark Kettenis  <kettenis@gnu.org>:
 
 	* config/i386/tm-i386.h (FLOAT_INFO): New define.
-	* i387-tdep.c (print_i387_value, print_i387_ext,
-	print_i387_status_word, print_i387_control_word, i387_float_info):
-	New functions, used to implement generic `info float' command.
+	
+	* i387-tdep.c (print_i387_value,print_i387_ext,i387_float_info):
+	  New functions, used to implement generic `info float' command.
 
 1999-12-06  Christopher Faylor  <cgf@cygnus.com>
 

[-- Attachment #4: ChangeLog-2001.diff --]
[-- Type: text/x-patch, Size: 557 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-2001	2002-01-06 05:37:53.000000000 +0100
+++ ./gdb-7.2.2/gdb/ChangeLog-2001	2011-04-14 15:55:22.631948002 +0200
@@ -8076,11 +8076,6 @@ Wed Apr  4 21:48:42 2001  Christopher Fa
 
 	* MAINTAINERS: Add myself to write after approval list.
 
-2001-03-09  Mark Kettenis  <kettenis@gnu.org>
-
-	* i387-tdep.c (print_i387_status_word): Fix printing of Stack
-	fault flag.  It is bit 6 of the x87 FPU status word, not bit 7.
-
 2001-03-08  Andrew Cagney  <ac131313@redhat.com>
 
 	* configure.host (m88*-harris-cxux*): Mark as obsolete.

[-- Attachment #5: ChangeLog-2002.diff --]
[-- Type: text/x-patch, Size: 1057 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-2002	2004-04-24 00:15:17.000000000 +0200
+++ ./gdb-7.2.2/gdb/ChangeLog-2002	2011-04-16 22:54:05.451736001 +0200
@@ -5396,12 +5396,14 @@ Fri Sep 13 07:42:09 2002  Andrew Cagney
 	* config/i386/tm-i386.h: Don't include "regcache.h".
 
 	* i387-tdep.h (i387_print_float_info): New prototype.
-	* i387-tdep.c (print_i387_value, print_i387_ext,
-	print_i387_status_word, print_i387_control_word): Add `struct
-	ui_file *' argument and use it for output.
-	(i387_print_float_info): Renamed from i387_float_info.  Add
-	`struct gdbarch *' and `struct ui_file *' arguments and use the
-	latter for output.
+
+	* i387-tdep.c (print_i387_value,print_i387_ext):
+	  Add `struct ui_file *' argument and use it for output.
+
+	* i387-tdep.c (i387_print_float_info):
+	  Renamed from i387_float_info.  Add `struct gdbarch *' and `struct ui_file *' arguments
+	  and use the latter for output.
+
 	* i386-tdep.c: Include "i387-tdep.h".
 	(i386_gdbarch_init): Set print_float_info.
 	* config/i386/tm-i386.h (i387_float_info): Remove prototype.

[-- Attachment #6: ChangeLog-2004.diff --]
[-- Type: text/x-patch, Size: 628 bytes --]

--- ./gdb-7.2/gdb/ChangeLog-2004	2006-10-24 22:23:56.000000000 +0200
+++ ./gdb-7.2.2/gdb/ChangeLog-2004	2011-04-16 22:54:31.555736002 +0200
@@ -3468,9 +3468,9 @@
 	* ui-out.c (ui_out_field_core_addr): Ditto.
 	* breakpoint.c (breakpoint_adjustment_warning): Ditto.
 	* exec.c (print_section_info): Ditto.
-	* i387-tdep.c (print_i387_status_word): Ditto.
-	(print_i387_control_word): Ditto.
-	(i387_print_float_info): Ditto.
+
+	* i387-tdep.c (i387_print_float_info): Ditto.
+
 	* maint.c (maint_print_section_info): Ditto.
 	* solib.c (info_sharedlibrary_command): Ditto.
 	* somsolib.c (som_sharedlibrary_info_command): Ditto.

[-- Attachment #7: i387-tdep.c.diff --]
[-- Type: text/x-patch, Size: 8147 bytes --]

--- ./gdb-7.2/gdb/i387-tdep.c	2010-05-05 21:30:36.000000000 +0200
+++ ./gdb-7.2.2/gdb/i387-tdep.c	2011-04-19 20:42:56.691027998 +0200
@@ -36,6 +36,106 @@
 #include "i387-tdep.h"
 #include "i386-xstate.h"
 
+struct
+{
+	const char *tag;
+	const char *description;
+}
+
+static const npx_exception_flags[6] =
+
+{
+	{"IE", "invalid operation"},
+	{"DE", "denormalized operand"},
+	{"ZE", "zero divide"},
+	{"OE", "overflow"},
+	{"UE", "underflow"},
+	{"PE", "precision"}
+};
+
+struct
+{
+	const char *tag;
+	const char *description;
+}
+
+static const npx_exception_mask[6] =
+
+{
+	{"IM", "invalid operation"},
+	{"DM", "denormalized operand"},
+	{"ZM", "zero divide"},
+	{"OM", "overflow"},
+	{"UM", "underflow"},
+	{"PM", "precision"}
+};
+
+static const char *npx_precision_control[4] = 
+{
+	"24-bits (single precision)", 
+	"(reserved)", 
+	"53-bits (double precision)", 
+	"64-bits (extended precision)"
+};
+
+static const char *npx_rounding_control[4] = 
+{
+	"round to nearest or even", 
+	"round down (towards -inf)", 
+	"round up (toward +inf)", 
+	"chop (truncate toward zero)"
+};
+
+/* print the numeric coprocessor extension (npx) status word */
+void print_npx_status_word(uint16_t npx_status_word, struct ui_file *file)
+{
+	fprintf_filtered(file, "status word              : 0x%04X\n", npx_status_word);
+
+	fprintf_filtered(file, "  exception flags        : ");
+
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0005) & 0x0001) ? npx_exception_flags[5].tag : "  ");	// precision
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0004) & 0x0001) ? npx_exception_flags[4].tag : "  ");	// underflow
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0003) & 0x0001) ? npx_exception_flags[3].tag : "  ");	// overflow
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0002) & 0x0001) ? npx_exception_flags[2].tag : "  ");	// zero divide
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0001) & 0x0001) ? npx_exception_flags[1].tag : "  ");	// denormalized operand
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0000) & 0x0001) ? npx_exception_flags[0].tag : "  ");	// invalid operation
+	fprintf_filtered(file, "\n");
+
+	fprintf_filtered(file, "  stack fault            : %u\n", ((npx_status_word >> 0x0006) & 0x0001));
+	fprintf_filtered(file, "  error summary status   : %u\n", ((npx_status_word >> 0x0007) & 0x0001));
+	fprintf_filtered(file, "  busy                   : %u\n", ((npx_status_word >> 0x000F) & 0x0001));
+	fprintf_filtered(file, "  top of stack pointer   : %u\n", ((npx_status_word >> 0x000B) & 0x0007));
+
+	fprintf_filtered(file, "  condition code         : ");
+
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x000E) & 0x0001) ? "C3" : "  ");
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x000A) & 0x0001) ? "C2" : "  ");
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0009) & 0x0001) ? "C1" : "  ");
+	fprintf_filtered(file, "%s ", ((npx_status_word >> 0x0008) & 0x0001) ? "C0" : "  ");
+	fprintf_filtered(file, "\n");
+}
+
+/* print the numeric coprocessor extension (npx) control word */
+void print_npx_control_word(uint16_t npx_control_word, struct ui_file *file)
+{
+	fprintf_filtered(file, "control word             : 0x%04X\n", npx_control_word);
+
+	fprintf_filtered(file, "  exception mask         : ");
+
+	fprintf_filtered(file, "%s ", ((npx_control_word >> 0x0005) & 0x0001) ? npx_exception_mask[5].tag : "  ");	// precision
+	fprintf_filtered(file, "%s ", ((npx_control_word >> 0x0004) & 0x0001) ? npx_exception_mask[4].tag : "  ");	// underflow
+	fprintf_filtered(file, "%s ", ((npx_control_word >> 0x0003) & 0x0001) ? npx_exception_mask[3].tag : "  ");	// overflow
+	fprintf_filtered(file, "%s ", ((npx_control_word >> 0x0002) & 0x0001) ? npx_exception_mask[2].tag : "  ");	// zero divide
+	fprintf_filtered(file, "%s ", ((npx_control_word >> 0x0001) & 0x0001) ? npx_exception_mask[1].tag : "  ");	// denormalized operand
+	fprintf_filtered(file, "%s ", ((npx_control_word >> 0x0000) & 0x0001) ? npx_exception_mask[0].tag : "  ");	// invalid operation
+	fprintf_filtered(file, "\n");
+
+	fprintf_filtered(file, "  precision control (PC) : %s\n", npx_precision_control[((npx_control_word >> 0x0008) & 0x0003)]);
+	fprintf_filtered(file, "  rounding control (RC)  : %s\n", npx_rounding_control[((npx_control_word >> 0x000A) & 0x0003)]);
+
+	fprintf_filtered(file, "  infinity control       : %u\n", ((npx_control_word >> 0x000C) & 0x0001));
+}
+
 /* Print the floating point number specified by RAW.  */
 
 static void
@@ -114,88 +214,6 @@ print_i387_ext (struct gdbarch *gdbarch,
     fputs_filtered (" Unsupported", file);
 }
 
-/* Print the status word STATUS.  */
-
-static void
-print_i387_status_word (unsigned int status, struct ui_file *file)
-{
-  fprintf_filtered (file, "Status Word:         %s",
-		    hex_string_custom (status, 4));
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0001) ? "IE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0002) ? "DE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0004) ? "ZE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0008) ? "OE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0010) ? "UE" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0020) ? "PE" : "  ");
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0080) ? "ES" : "  ");
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0040) ? "SF" : "  ");
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (status & 0x0100) ? "C0" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0200) ? "C1" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x0400) ? "C2" : "  ");
-  fprintf_filtered (file, " %s", (status & 0x4000) ? "C3" : "  ");
-
-  fputs_filtered ("\n", file);
-
-  fprintf_filtered (file,
-		    "                       TOP: %d\n", ((status >> 11) & 7));
-}
-
-/* Print the control word CONTROL.  */
-
-static void
-print_i387_control_word (unsigned int control, struct ui_file *file)
-{
-  fprintf_filtered (file, "Control Word:        %s",
-		    hex_string_custom (control, 4));
-  fputs_filtered ("  ", file);
-  fprintf_filtered (file, " %s", (control & 0x0001) ? "IM" : "  ");
-  fprintf_filtered (file, " %s", (control & 0x0002) ? "DM" : "  ");
-  fprintf_filtered (file, " %s", (control & 0x0004) ? "ZM" : "  ");
-  fprintf_filtered (file, " %s", (control & 0x0008) ? "OM" : "  ");
-  fprintf_filtered (file, " %s", (control & 0x0010) ? "UM" : "  ");
-  fprintf_filtered (file, " %s", (control & 0x0020) ? "PM" : "  ");
-
-  fputs_filtered ("\n", file);
-
-  fputs_filtered ("                       PC: ", file);
-  switch ((control >> 8) & 3)
-    {
-    case 0:
-      fputs_filtered ("Single Precision (24-bits)\n", file);
-      break;
-    case 1:
-      fputs_filtered ("Reserved\n", file);
-      break;
-    case 2:
-      fputs_filtered ("Double Precision (53-bits)\n", file);
-      break;
-    case 3:
-      fputs_filtered ("Extended Precision (64-bits)\n", file);
-      break;
-    }
-      
-  fputs_filtered ("                       RC: ", file);
-  switch ((control >> 10) & 3)
-    {
-    case 0:
-      fputs_filtered ("Round to nearest\n", file);
-      break;
-    case 1:
-      fputs_filtered ("Round down\n", file);
-      break;
-    case 2:
-      fputs_filtered ("Round up\n", file);
-      break;
-    case 3:
-      fputs_filtered ("Round toward zero\n", file);
-      break;
-    }
-}
-
 /* Print out the i387 floating point state.  Note that we ignore FRAME
    in the code below.  That's OK since floating-point registers are
    never saved on the stack.  */
@@ -268,8 +286,12 @@ i387_print_float_info (struct gdbarch *g
 
   fputs_filtered ("\n", file);
 
-  print_i387_status_word (fstat, file);
-  print_i387_control_word (fctrl, file);
+	print_npx_status_word(fstat, file);
+	fprintf_filtered(file, "\n");
+
+	print_npx_control_word(fctrl, file);
+	fprintf_filtered(file, "\n");
+
   fprintf_filtered (file, "Tag Word:            %s\n",
 		    hex_string_custom (ftag, 4));
   fprintf_filtered (file, "Instruction Pointer: %s:",

Re: print_npx_status_word

[Pedro Alves]

Please, do read src/gdb/CONTRIBUTE.  All of it.

It's okay to not know the standards by heart, but we won't
put code in the tree that doesn't follow the standards.
No exceptions.

-- 
Pedro Alves

Re: print_npx_status_word

[Tom Tromey]

Tom> As others have noted, ChangeLog is "prepend only".
Tom> We generally don't edit old entries.

Michael> As I stated before, it would be unwise though to keep unmaintained
Michael> function names in the log entries. You can apply the changes
Michael> optionally, but I strongly advise against retaining the original log
Michael> entries.

The log is just that, a log.  Think of it as a primitive version of VC
commit messages.  It is a document describe a sequence of changes, not a
document describing the current state of the source -- the source itself
plays that role.

Michael> If you don't like the indentation, I suggest you instead modify
Michael> you tab-settings locally. Personally I'm used programming with
Michael> a little more space than.

The usual rule in all projects is "consistent is better than good".
GDB fixed the basics of its style many years ago, for better or worse.
So, please conform to the prevailing style.

Tom> No `//' comments.

Michael> I have taken this into account, does that also hold for inline
Michael> comments?

Yes.  GDB is mostly on C89.  This is fairly typical in the GNU tools.

Tom

print_npx_status_word

[Baars, M.J.]

[-- Attachment #1: Type: text/plain, Size: 62 bytes --]

hi all,

who will apply this patch for me?

regards,
michael


[-- Attachment #2: patches-gdb-7.2.tar.gz --]
[-- Type: application/x-gzip, Size: 1611 bytes --]