* RFA: Fix target/44750
@ 2010-11-04 18:28 Joern Rennecke
2010-11-04 20:43 ` Paul Koning
[not found] ` <76BC23E8-E889-405A-9CBD-651F4D63A4B6@dell.com>
0 siblings, 2 replies; 5+ messages in thread
From: Joern Rennecke @ 2010-11-04 18:28 UTC (permalink / raw)
To: gcc-patches; +Cc: ni1d
[-- Attachment #1: Type: text/plain, Size: 195 bytes --]
This patch fixes the remaining issues for the --enable-werror-always build
of pdp11.
tested by building 'all-gcc' for i686-pc-linux-gnu X pdp11-elf
using gcc (GCC) 4.6.0 20101103 (experimental).
[-- Attachment #2: pr44750-patch --]
[-- Type: text/plain, Size: 2975 bytes --]
2010-11-04 Joern Rennecke <amylaar@spamcop.net>
PR target/44750
* config/pdp11/pdp11.c: (pdp11_secondary_reload): Now static.
(pdp11_preferred_reload_class): Rename class to rclass.
(pdp11_preferred_output_reload_class): Likewise.
* config/pdp11/t-pdp11 (dwarf2out.o): Add -Wno-error=type-limits to
ALL_COMPILERFLAGS .
(java/constants.o): Add -Wno-error to ALL_COMPILERFLAGS.
Index: config/pdp11/pdp11.c
===================================================================
--- config/pdp11/pdp11.c (revision 166313)
+++ config/pdp11/pdp11.c (working copy)
@@ -1630,18 +1630,18 @@ pdp11_cannot_change_mode_class (enum mac
loading is easier into LOAD_FPU_REGS than FPU_REGS! */
static reg_class_t
-pdp11_preferred_reload_class (rtx x, reg_class_t class)
+pdp11_preferred_reload_class (rtx x, reg_class_t rclass)
{
- if (class == FPU_REGS)
+ if (rclass == FPU_REGS)
return LOAD_FPU_REGS;
- if (class == ALL_REGS)
+ if (rclass == ALL_REGS)
{
if (FLOAT_MODE_P (GET_MODE (x)))
return LOAD_FPU_REGS;
else
return GENERAL_REGS;
}
- return class;
+ return rclass;
}
/* TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
@@ -1654,18 +1654,18 @@ pdp11_preferred_reload_class (rtx x, reg
loading is easier into LOAD_FPU_REGS than FPU_REGS! */
static reg_class_t
-pdp11_preferred_output_reload_class (rtx x, reg_class_t class)
+pdp11_preferred_output_reload_class (rtx x, reg_class_t rclass)
{
- if (class == FPU_REGS)
+ if (rclass == FPU_REGS)
return LOAD_FPU_REGS;
- if (class == ALL_REGS)
+ if (rclass == ALL_REGS)
{
if (FLOAT_MODE_P (GET_MODE (x)))
return LOAD_FPU_REGS;
else
return GENERAL_REGS;
}
- return class;
+ return rclass;
}
@@ -1674,7 +1674,7 @@ pdp11_preferred_output_reload_class (rtx
FPU registers AC4 and AC5 (class NO_LOAD_FPU_REGS) require an
intermediate register (AC0-AC3: LOAD_FPU_REGS). Everything else
can be loade/stored directly. */
-reg_class_t
+static reg_class_t
pdp11_secondary_reload (bool in_p ATTRIBUTE_UNUSED,
rtx x,
reg_class_t reload_class,
Index: config/pdp11/t-pdp11
===================================================================
--- config/pdp11/t-pdp11 (revision 166313)
+++ config/pdp11/t-pdp11 (working copy)
@@ -34,3 +34,13 @@ dp-bit.c: $(srcdir)/config/fp-bit.c
cat $(srcdir)/config/fp-bit.c > dp-bit.c
MULTILIB_OPTIONS = msoft-float
+
+# Because the pdp11 POINTER_SIZE is only 16, in dwarf2out.c,
+# DWARF_ARANGES_PAD_SIZE is 0, thus a loop in output_aranges that checks
+# (i < (unsigned) DWARF_ARANGES_PAD_SIZE) elicits a warning that the
+# comparison is always false.
+dwarf2out.o: ALL_COMPILERFLAGS += -Wno-error=type-limits
+
+# Likewise, java/constants.c:build_constants_constructor has a negative shift
+# count (in never-executed code) due to the small POINTER_SIZE.
+java/constants.o: ALL_COMPILERFLAGS += -Wno-error
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: RFA: Fix target/44750
2010-11-04 18:28 RFA: Fix target/44750 Joern Rennecke
@ 2010-11-04 20:43 ` Paul Koning
2010-11-04 22:50 ` Paul Koning
[not found] ` <76BC23E8-E889-405A-9CBD-651F4D63A4B6@dell.com>
1 sibling, 1 reply; 5+ messages in thread
From: Paul Koning @ 2010-11-04 20:43 UTC (permalink / raw)
To: Joern Rennecke; +Cc: gcc-patches
Ok -- thanks!
paul
On Nov 4, 2010, at 2:25 PM, Joern Rennecke wrote:
> This patch fixes the remaining issues for the --enable-werror-always build
> of pdp11.
> tested by building 'all-gcc' for i686-pc-linux-gnu X pdp11-elf
> using gcc (GCC) 4.6.0 20101103 (experimental).
> 2010-11-04 Joern Rennecke <amylaar@spamcop.net>
>
> PR target/44750
> * config/pdp11/pdp11.c: (pdp11_secondary_reload): Now static.
> (pdp11_preferred_reload_class): Rename class to rclass.
> (pdp11_preferred_output_reload_class): Likewise.
> * config/pdp11/t-pdp11 (dwarf2out.o): Add -Wno-error=type-limits to
> ALL_COMPILERFLAGS .
> (java/constants.o): Add -Wno-error to ALL_COMPILERFLAGS.
>
> Index: config/pdp11/pdp11.c
> ===================================================================
> --- config/pdp11/pdp11.c (revision 166313)
> +++ config/pdp11/pdp11.c (working copy)
> @@ -1630,18 +1630,18 @@ pdp11_cannot_change_mode_class (enum mac
> loading is easier into LOAD_FPU_REGS than FPU_REGS! */
>
> static reg_class_t
> -pdp11_preferred_reload_class (rtx x, reg_class_t class)
> +pdp11_preferred_reload_class (rtx x, reg_class_t rclass)
> {
> - if (class == FPU_REGS)
> + if (rclass == FPU_REGS)
> return LOAD_FPU_REGS;
> - if (class == ALL_REGS)
> + if (rclass == ALL_REGS)
> {
> if (FLOAT_MODE_P (GET_MODE (x)))
> return LOAD_FPU_REGS;
> else
> return GENERAL_REGS;
> }
> - return class;
> + return rclass;
> }
>
> /* TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
> @@ -1654,18 +1654,18 @@ pdp11_preferred_reload_class (rtx x, reg
> loading is easier into LOAD_FPU_REGS than FPU_REGS! */
>
> static reg_class_t
> -pdp11_preferred_output_reload_class (rtx x, reg_class_t class)
> +pdp11_preferred_output_reload_class (rtx x, reg_class_t rclass)
> {
> - if (class == FPU_REGS)
> + if (rclass == FPU_REGS)
> return LOAD_FPU_REGS;
> - if (class == ALL_REGS)
> + if (rclass == ALL_REGS)
> {
> if (FLOAT_MODE_P (GET_MODE (x)))
> return LOAD_FPU_REGS;
> else
> return GENERAL_REGS;
> }
> - return class;
> + return rclass;
> }
>
>
> @@ -1674,7 +1674,7 @@ pdp11_preferred_output_reload_class (rtx
> FPU registers AC4 and AC5 (class NO_LOAD_FPU_REGS) require an
> intermediate register (AC0-AC3: LOAD_FPU_REGS). Everything else
> can be loade/stored directly. */
> -reg_class_t
> +static reg_class_t
> pdp11_secondary_reload (bool in_p ATTRIBUTE_UNUSED,
> rtx x,
> reg_class_t reload_class,
> Index: config/pdp11/t-pdp11
> ===================================================================
> --- config/pdp11/t-pdp11 (revision 166313)
> +++ config/pdp11/t-pdp11 (working copy)
> @@ -34,3 +34,13 @@ dp-bit.c: $(srcdir)/config/fp-bit.c
> cat $(srcdir)/config/fp-bit.c > dp-bit.c
>
> MULTILIB_OPTIONS = msoft-float
> +
> +# Because the pdp11 POINTER_SIZE is only 16, in dwarf2out.c,
> +# DWARF_ARANGES_PAD_SIZE is 0, thus a loop in output_aranges that checks
> +# (i < (unsigned) DWARF_ARANGES_PAD_SIZE) elicits a warning that the
> +# comparison is always false.
> +dwarf2out.o: ALL_COMPILERFLAGS += -Wno-error=type-limits
> +
> +# Likewise, java/constants.c:build_constants_constructor has a negative shift
> +# count (in never-executed code) due to the small POINTER_SIZE.
> +java/constants.o: ALL_COMPILERFLAGS += -Wno-error
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: RFA: Fix target/44750
2010-11-04 20:43 ` Paul Koning
@ 2010-11-04 22:50 ` Paul Koning
0 siblings, 0 replies; 5+ messages in thread
From: Paul Koning @ 2010-11-04 22:50 UTC (permalink / raw)
To: Joern Rennecke, gcc-patches
My apologies, I misread the patch. I can only approve the pdp11 part of this.
paul
On Nov 4, 2010, at 4:34 PM, Paul Koning wrote:
> Ok -- thanks!
>
> paul
>
> On Nov 4, 2010, at 2:25 PM, Joern Rennecke wrote:
>
>> This patch fixes the remaining issues for the --enable-werror-always build
>> of pdp11.
>> tested by building 'all-gcc' for i686-pc-linux-gnu X pdp11-elf
>> using gcc (GCC) 4.6.0 20101103 (experimental).
>> 2010-11-04 Joern Rennecke <amylaar@spamcop.net>
>>
>> PR target/44750
>> * config/pdp11/pdp11.c: (pdp11_secondary_reload): Now static.
>> (pdp11_preferred_reload_class): Rename class to rclass.
>> (pdp11_preferred_output_reload_class): Likewise.
>> * config/pdp11/t-pdp11 (dwarf2out.o): Add -Wno-error=type-limits to
>> ALL_COMPILERFLAGS .
>> (java/constants.o): Add -Wno-error to ALL_COMPILERFLAGS.
>>
>> Index: config/pdp11/pdp11.c
>> ===================================================================
>> --- config/pdp11/pdp11.c (revision 166313)
>> +++ config/pdp11/pdp11.c (working copy)
>> @@ -1630,18 +1630,18 @@ pdp11_cannot_change_mode_class (enum mac
>> loading is easier into LOAD_FPU_REGS than FPU_REGS! */
>>
>> static reg_class_t
>> -pdp11_preferred_reload_class (rtx x, reg_class_t class)
>> +pdp11_preferred_reload_class (rtx x, reg_class_t rclass)
>> {
>> - if (class == FPU_REGS)
>> + if (rclass == FPU_REGS)
>> return LOAD_FPU_REGS;
>> - if (class == ALL_REGS)
>> + if (rclass == ALL_REGS)
>> {
>> if (FLOAT_MODE_P (GET_MODE (x)))
>> return LOAD_FPU_REGS;
>> else
>> return GENERAL_REGS;
>> }
>> - return class;
>> + return rclass;
>> }
>>
>> /* TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
>> @@ -1654,18 +1654,18 @@ pdp11_preferred_reload_class (rtx x, reg
>> loading is easier into LOAD_FPU_REGS than FPU_REGS! */
>>
>> static reg_class_t
>> -pdp11_preferred_output_reload_class (rtx x, reg_class_t class)
>> +pdp11_preferred_output_reload_class (rtx x, reg_class_t rclass)
>> {
>> - if (class == FPU_REGS)
>> + if (rclass == FPU_REGS)
>> return LOAD_FPU_REGS;
>> - if (class == ALL_REGS)
>> + if (rclass == ALL_REGS)
>> {
>> if (FLOAT_MODE_P (GET_MODE (x)))
>> return LOAD_FPU_REGS;
>> else
>> return GENERAL_REGS;
>> }
>> - return class;
>> + return rclass;
>> }
>>
>>
>> @@ -1674,7 +1674,7 @@ pdp11_preferred_output_reload_class (rtx
>> FPU registers AC4 and AC5 (class NO_LOAD_FPU_REGS) require an
>> intermediate register (AC0-AC3: LOAD_FPU_REGS). Everything else
>> can be loade/stored directly. */
>> -reg_class_t
>> +static reg_class_t
>> pdp11_secondary_reload (bool in_p ATTRIBUTE_UNUSED,
>> rtx x,
>> reg_class_t reload_class,
>> Index: config/pdp11/t-pdp11
>> ===================================================================
>> --- config/pdp11/t-pdp11 (revision 166313)
>> +++ config/pdp11/t-pdp11 (working copy)
>> @@ -34,3 +34,13 @@ dp-bit.c: $(srcdir)/config/fp-bit.c
>> cat $(srcdir)/config/fp-bit.c > dp-bit.c
>>
>> MULTILIB_OPTIONS = msoft-float
>> +
>> +# Because the pdp11 POINTER_SIZE is only 16, in dwarf2out.c,
>> +# DWARF_ARANGES_PAD_SIZE is 0, thus a loop in output_aranges that checks
>> +# (i < (unsigned) DWARF_ARANGES_PAD_SIZE) elicits a warning that the
>> +# comparison is always false.
>> +dwarf2out.o: ALL_COMPILERFLAGS += -Wno-error=type-limits
>> +
>> +# Likewise, java/constants.c:build_constants_constructor has a negative shift
>> +# count (in never-executed code) due to the small POINTER_SIZE.
>> +java/constants.o: ALL_COMPILERFLAGS += -Wno-error
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: RFA: Fix target/44750
[not found] ` <76BC23E8-E889-405A-9CBD-651F4D63A4B6@dell.com>
@ 2010-11-05 4:53 ` Joern Rennecke
2010-11-05 10:55 ` Paul Koning
0 siblings, 1 reply; 5+ messages in thread
From: Joern Rennecke @ 2010-11-05 4:53 UTC (permalink / raw)
To: Paul Koning; +Cc: gcc-patches
[-- Attachment #1: Type: text/plain, Size: 421 bytes --]
Quoting Paul Koning <paul_koning@dell.com>:
> Joern,
>
> I ran into a problem with this.
Oops, while we can rely on a fairly recent version of GNU make, the same
is not true of the bootstrap compiler. So there is still a good reason
to stick to the *.o-warn mechanism.
The attached patch makes use of this so that we use just -Wno-error, and only
when we're actually using -Werror in the first place.
OK to apply?
[-- Attachment #2: pdp11-warn-diff --]
[-- Type: text/plain, Size: 1060 bytes --]
2010-11-05 Joern Rennecke <amylaar@spamcop.net>
PR target/44750
* config/pdp11/t-pdp11 (dwarf2out.o, java/constants.o): Undo last
change. Set $@-warn.o to -Wno-error.
Index: config/pdp11/t-pdp11
===================================================================
--- config/pdp11/t-pdp11 (revision 166331)
+++ config/pdp11/t-pdp11 (working copy)
@@ -39,8 +39,10 @@ MULTILIB_OPTIONS = msoft-float
# DWARF_ARANGES_PAD_SIZE is 0, thus a loop in output_aranges that checks
# (i < (unsigned) DWARF_ARANGES_PAD_SIZE) elicits a warning that the
# comparison is always false.
-dwarf2out.o: ALL_COMPILERFLAGS += -Wno-error=type-limits
+# We could say "-Werror -Wno-error=type-limits", alas, not all supported
+# gcc bootstrap compilers support the latter option.
+dwarf2out.o-warn = -Wno-error
# Likewise, java/constants.c:build_constants_constructor has a negative shift
# count (in never-executed code) due to the small POINTER_SIZE.
-java/constants.o: ALL_COMPILERFLAGS += -Wno-error
+java/constants.o-warn = -Wno-error
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: RFA: Fix target/44750
2010-11-05 4:53 ` Joern Rennecke
@ 2010-11-05 10:55 ` Paul Koning
0 siblings, 0 replies; 5+ messages in thread
From: Paul Koning @ 2010-11-05 10:55 UTC (permalink / raw)
To: Joern Rennecke; +Cc: gcc-patches
Ok. Thanks!
paul
On Nov 5, 2010, at 12:35 AM, Joern Rennecke wrote:
> Quoting Paul Koning <paul_koning@dell.com>:
>
>> Joern,
>>
>> I ran into a problem with this.
>
> Oops, while we can rely on a fairly recent version of GNU make, the same
> is not true of the bootstrap compiler. So there is still a good reason
> to stick to the *.o-warn mechanism.
>
> The attached patch makes use of this so that we use just -Wno-error, and only
> when we're actually using -Werror in the first place.
>
> OK to apply?
>
>
> 2010-11-05 Joern Rennecke <amylaar@spamcop.net>
>
> PR target/44750
> * config/pdp11/t-pdp11 (dwarf2out.o, java/constants.o): Undo last
> change. Set $@-warn.o to -Wno-error.
>
> Index: config/pdp11/t-pdp11
> ===================================================================
> --- config/pdp11/t-pdp11 (revision 166331)
> +++ config/pdp11/t-pdp11 (working copy)
> @@ -39,8 +39,10 @@ MULTILIB_OPTIONS = msoft-float
> # DWARF_ARANGES_PAD_SIZE is 0, thus a loop in output_aranges that checks
> # (i < (unsigned) DWARF_ARANGES_PAD_SIZE) elicits a warning that the
> # comparison is always false.
> -dwarf2out.o: ALL_COMPILERFLAGS += -Wno-error=type-limits
> +# We could say "-Werror -Wno-error=type-limits", alas, not all supported
> +# gcc bootstrap compilers support the latter option.
> +dwarf2out.o-warn = -Wno-error
>
> # Likewise, java/constants.c:build_constants_constructor has a negative shift
> # count (in never-executed code) due to the small POINTER_SIZE.
> -java/constants.o: ALL_COMPILERFLAGS += -Wno-error
> +java/constants.o-warn = -Wno-error
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2010-11-05 10:45 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-11-04 18:28 RFA: Fix target/44750 Joern Rennecke
2010-11-04 20:43 ` Paul Koning
2010-11-04 22:50 ` Paul Koning
[not found] ` <76BC23E8-E889-405A-9CBD-651F4D63A4B6@dell.com>
2010-11-05 4:53 ` Joern Rennecke
2010-11-05 10:55 ` Paul Koning
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).