* [PATCH] pair-fusion: fix for older GCC
@ 2024-06-03 7:57 Marc Poulhiès
2024-06-03 8:08 ` Richard Sandiford
0 siblings, 1 reply; 3+ messages in thread
From: Marc Poulhiès @ 2024-06-03 7:57 UTC (permalink / raw)
To: gcc-patches; +Cc: Ajit Agarwal, Richard Sandiford, Marc Poulhiès
Older GCCs fail with:
.../gcc/pair-fusion.cc: In member function ‘bool pair_fusion_bb_info::fuse_pair(bool, unsigned int, int, rtl_ssa::insn_info*, rtl_ssa::in
sn_info*, base_cand&, const rtl_ssa::insn_range_info&)’:
.../gcc/pair-fusion.cc:1790:40: error: ‘writeback’ is not a class, namespace, or enumeration
if (m_pass->should_handle_writeback (writeback::ALL)
Renaming the enum type works around the name conflict with the local
variable and also prevents future similar conflicts.
gcc/ChangeLog:
* pair-fusion.h (enum class writeback): Rename to...
(enum class writeback_type): ...this.
(struct pair_fusion): Adjust type name after renaming.
* pair-fusion.cc (pair_fusion_bb_info::track_access): Likewise.
(pair_fusion_bb_info::fuse_pair): Likewise.
(pair_fusion::process_block): Likewise.
---
Patch discussed in https://inbox.sourceware.org/gcc-patches/mptwmn93njq.fsf@arm.com/
Tested on x86_64-linux-gnu. OK for master?
gcc/pair-fusion.cc | 6 +++---
gcc/pair-fusion.h | 4 ++--
2 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/gcc/pair-fusion.cc b/gcc/pair-fusion.cc
index 9f897ac04e2..26b2284ed37 100644
--- a/gcc/pair-fusion.cc
+++ b/gcc/pair-fusion.cc
@@ -426,7 +426,7 @@ pair_fusion_bb_info::track_access (insn_info *insn, bool load_p, rtx mem)
return;
// Ignore writeback accesses if the hook says to do so.
- if (!m_pass->should_handle_writeback (writeback::EXISTING)
+ if (!m_pass->should_handle_writeback (writeback_type::EXISTING)
&& GET_RTX_CLASS (GET_CODE (XEXP (mem, 0))) == RTX_AUTOINC)
return;
@@ -1787,7 +1787,7 @@ pair_fusion_bb_info::fuse_pair (bool load_p,
// update of the base register and try and fold it in to make this into a
// writeback pair.
insn_info *trailing_add = nullptr;
- if (m_pass->should_handle_writeback (writeback::ALL)
+ if (m_pass->should_handle_writeback (writeback_type::ALL)
&& !writeback_effect
&& (!load_p || (!refers_to_regno_p (base_regno, base_regno + 1,
XEXP (pats[0], 0), nullptr)
@@ -2996,7 +2996,7 @@ void pair_fusion::process_block (bb_info *bb)
rtx pat = PATTERN (rti);
bool load_p;
if (reload_completed
- && should_handle_writeback (writeback::ALL)
+ && should_handle_writeback (writeback_type::ALL)
&& pair_mem_insn_p (rti, load_p))
try_promote_writeback (insn, load_p);
diff --git a/gcc/pair-fusion.h b/gcc/pair-fusion.h
index 2a38dc8f743..45e4edceecb 100644
--- a/gcc/pair-fusion.h
+++ b/gcc/pair-fusion.h
@@ -75,7 +75,7 @@ struct alias_walker;
// When querying should_handle_writeback, this enum is used to
// qualify which opportunities we are asking about.
-enum class writeback {
+enum class writeback_type {
// Only those writeback opportunities that arise from existing
// auto-increment accesses.
EXISTING,
@@ -123,7 +123,7 @@ struct pair_fusion {
// Return true if we should try to handle writeback opportunities.
// WHICH determines the kinds of writeback opportunities the caller
// is asking about.
- virtual bool should_handle_writeback (enum writeback which) = 0;
+ virtual bool should_handle_writeback (writeback_type which) = 0;
// Given BASE_MEM, the mem from the lower candidate access for a pair,
// and LOAD_P (true if the access is a load), check if we should proceed
--
2.45.1
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] pair-fusion: fix for older GCC
2024-06-03 7:57 [PATCH] pair-fusion: fix for older GCC Marc Poulhiès
@ 2024-06-03 8:08 ` Richard Sandiford
2024-06-03 9:21 ` Marc Poulhiès
0 siblings, 1 reply; 3+ messages in thread
From: Richard Sandiford @ 2024-06-03 8:08 UTC (permalink / raw)
To: Marc Poulhiès; +Cc: gcc-patches, Ajit Agarwal
Marc Poulhiès <poulhies@adacore.com> writes:
> Older GCCs fail with:
>
> .../gcc/pair-fusion.cc: In member function ‘bool pair_fusion_bb_info::fuse_pair(bool, unsigned int, int, rtl_ssa::insn_info*, rtl_ssa::in
> sn_info*, base_cand&, const rtl_ssa::insn_range_info&)’:
> .../gcc/pair-fusion.cc:1790:40: error: ‘writeback’ is not a class, namespace, or enumeration
> if (m_pass->should_handle_writeback (writeback::ALL)
>
> Renaming the enum type works around the name conflict with the local
> variable and also prevents future similar conflicts.
>
> gcc/ChangeLog:
>
> * pair-fusion.h (enum class writeback): Rename to...
> (enum class writeback_type): ...this.
> (struct pair_fusion): Adjust type name after renaming.
> * pair-fusion.cc (pair_fusion_bb_info::track_access): Likewise.
> (pair_fusion_bb_info::fuse_pair): Likewise.
> (pair_fusion::process_block): Likewise.
OK, thanks, and sorry for missing this during the review.
Richard
> ---
> Patch discussed in https://inbox.sourceware.org/gcc-patches/mptwmn93njq.fsf@arm.com/
>
> Tested on x86_64-linux-gnu. OK for master?
>
> gcc/pair-fusion.cc | 6 +++---
> gcc/pair-fusion.h | 4 ++--
> 2 files changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/gcc/pair-fusion.cc b/gcc/pair-fusion.cc
> index 9f897ac04e2..26b2284ed37 100644
> --- a/gcc/pair-fusion.cc
> +++ b/gcc/pair-fusion.cc
> @@ -426,7 +426,7 @@ pair_fusion_bb_info::track_access (insn_info *insn, bool load_p, rtx mem)
> return;
>
> // Ignore writeback accesses if the hook says to do so.
> - if (!m_pass->should_handle_writeback (writeback::EXISTING)
> + if (!m_pass->should_handle_writeback (writeback_type::EXISTING)
> && GET_RTX_CLASS (GET_CODE (XEXP (mem, 0))) == RTX_AUTOINC)
> return;
>
> @@ -1787,7 +1787,7 @@ pair_fusion_bb_info::fuse_pair (bool load_p,
> // update of the base register and try and fold it in to make this into a
> // writeback pair.
> insn_info *trailing_add = nullptr;
> - if (m_pass->should_handle_writeback (writeback::ALL)
> + if (m_pass->should_handle_writeback (writeback_type::ALL)
> && !writeback_effect
> && (!load_p || (!refers_to_regno_p (base_regno, base_regno + 1,
> XEXP (pats[0], 0), nullptr)
> @@ -2996,7 +2996,7 @@ void pair_fusion::process_block (bb_info *bb)
> rtx pat = PATTERN (rti);
> bool load_p;
> if (reload_completed
> - && should_handle_writeback (writeback::ALL)
> + && should_handle_writeback (writeback_type::ALL)
> && pair_mem_insn_p (rti, load_p))
> try_promote_writeback (insn, load_p);
>
> diff --git a/gcc/pair-fusion.h b/gcc/pair-fusion.h
> index 2a38dc8f743..45e4edceecb 100644
> --- a/gcc/pair-fusion.h
> +++ b/gcc/pair-fusion.h
> @@ -75,7 +75,7 @@ struct alias_walker;
>
> // When querying should_handle_writeback, this enum is used to
> // qualify which opportunities we are asking about.
> -enum class writeback {
> +enum class writeback_type {
> // Only those writeback opportunities that arise from existing
> // auto-increment accesses.
> EXISTING,
> @@ -123,7 +123,7 @@ struct pair_fusion {
> // Return true if we should try to handle writeback opportunities.
> // WHICH determines the kinds of writeback opportunities the caller
> // is asking about.
> - virtual bool should_handle_writeback (enum writeback which) = 0;
> + virtual bool should_handle_writeback (writeback_type which) = 0;
>
> // Given BASE_MEM, the mem from the lower candidate access for a pair,
> // and LOAD_P (true if the access is a load), check if we should proceed
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] pair-fusion: fix for older GCC
2024-06-03 8:08 ` Richard Sandiford
@ 2024-06-03 9:21 ` Marc Poulhiès
0 siblings, 0 replies; 3+ messages in thread
From: Marc Poulhiès @ 2024-06-03 9:21 UTC (permalink / raw)
To: gcc-patches; +Cc: Ajit Agarwal, richard.sandiford
Richard Sandiford <richard.sandiford@arm.com> writes:
> Marc Poulhiès <poulhies@adacore.com> writes:
>> Older GCCs fail with:
>>
>> .../gcc/pair-fusion.cc: In member function ‘bool pair_fusion_bb_info::fuse_pair(bool, unsigned int, int, rtl_ssa::insn_info*, rtl_ssa::in
>> sn_info*, base_cand&, const rtl_ssa::insn_range_info&)’:
>> .../gcc/pair-fusion.cc:1790:40: error: ‘writeback’ is not a class, namespace, or enumeration
>> if (m_pass->should_handle_writeback (writeback::ALL)
>>
>> Renaming the enum type works around the name conflict with the local
>> variable and also prevents future similar conflicts.
>>
>> gcc/ChangeLog:
>>
>> * pair-fusion.h (enum class writeback): Rename to...
>> (enum class writeback_type): ...this.
>> (struct pair_fusion): Adjust type name after renaming.
>> * pair-fusion.cc (pair_fusion_bb_info::track_access): Likewise.
>> (pair_fusion_bb_info::fuse_pair): Likewise.
>> (pair_fusion::process_block): Likewise.
>
> OK, thanks, and sorry for missing this during the review.
This breaks aarch64:
https://builder.sourceware.org/buildbot/#/builders/266/builds/3487/steps/4/logs/stdio
I'll try to send a followup fix quickly...
Marc
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-06-03 9:21 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-06-03 7:57 [PATCH] pair-fusion: fix for older GCC Marc Poulhiès
2024-06-03 8:08 ` Richard Sandiford
2024-06-03 9:21 ` Marc Poulhiès
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).