From: Richard Biener <rguenther@suse.de>
To: Jeff Law <law@redhat.com>,Jiufu Guo <guojiufu@linux.ibm.com>
Cc: gcc-patches@gcc.gnu.org,Jakub Jelinek <jakub@redhat.com>,Daniel
Berlin
<dberlin@dberlin.org>,segher@kernel.crashing.org,wschmidt@linux.ibm.com
Subject: Re: [PATCH] A jump threading opportunity for condition branch
Date: Thu, 30 May 2019 06:41:00 -0000 [thread overview]
Message-ID: <63105C16-2CE6-417D-A99A-7C13B45378B2@suse.de> (raw)
In-Reply-To: <18b7e6a9-ee69-434d-adbf-75661113f1f6@redhat.com>
On May 29, 2019 10:12:31 PM GMT+02:00, Jeff Law <law@redhat.com> wrote:
>On 5/23/19 6:05 AM, Jiufu Guo wrote:
>> Hi,
>>
>> Richard Biener <rguenther@suse.de> writes:
>>
>>> On Tue, 21 May 2019, Jiufu Guo wrote:
>>>
>
>>>>
>>>> +/* Return true if PHI's INDEX-th incoming value is a CMP, and the
>CMP is
>>>> + defined in the incoming basic block. Otherwise return false.
>*/
>>>> +static bool
>>>> +cmp_from_unconditional_block (gphi *phi, int index)
>>>> +{
>>>> + tree value = gimple_phi_arg_def (phi, index);
>>>> + if (!(TREE_CODE (value) == SSA_NAME && has_single_use (value)))
>>>> + return false;
>>> Not sure why we should reject a constant here but I guess we
>>> expect it to find a simplified condition anyways ;)
>>>
>> Const could be accepted here, like "# t_9 = PHI <5(3), t_17(4)>". I
>> found this case is already handled by other jump-threading code, like
>> 'ethread' pass.
>Right. There's no need to handle constants here. They'll result in
>trivially discoverable jump threading opportunities.
>
>>>> + /* Check if phi's incoming value is defined in the incoming
>basic_block. */
>>>> + edge e = gimple_phi_arg_edge (phi, index);
>>>> + if (def->bb != e->src)
>>>> + return false;
>>> why does this matter?
>>>
>> Through preparing pathes and duplicating block, this transform can
>also
>> help to combine a cmp in previous block and a gcond in current block.
>> "if (def->bb != e->src)" make sure the cmp is define in the incoming
>> block of the current; and then combining "cmp with gcond" is safe.
>If
>> the cmp is defined far from the incoming block, it would be hard to
>> achieve the combining, and the transform may not needed.
>I don't think it's strictly needed in the long term and could be
>addressed in a follow-up if we can find cases where it helps. I think
>we'd just need to double check insertion of the new conditional branch
>to relax this if we cared.
>
>However, I would expect sinking to have done is job here and would be
>surprised if trying to handle this actually improved any real world
>code.
>>
>>>> +
>>>> + if (!single_succ_p (def->bb))
>>>> + return false;
>>> Or this? The actual threading will ensure this will hold true.
>>>
>> Yes, other thread code check this and ensure it to be true, like
>> function thread_through_normal_block. Since this new function is
>invoked
>> outside thread_through_normal_block, so, checking single_succ_p is
>also
>> needed for this case.
>Agreed that it's needed. Consider if the source block has multiple
>successors. Where do we insert the copy of the conditional branch?
We're duplicating its block? That is, we are isolating a path into a conditional - that's always possible? I wanted to make sure that when threading threads through a conditional in the block with the compare we'd add the extra tail duplication? AFAIK we're still looking at unmodified CFG here?
>
>>>> +{
>>>> + gimple *gs = last_and_only_stmt (bb);
>>>> + if (gs == NULL)
>>>> + return false;
>>>> +
>>>> + if (gimple_code (gs) != GIMPLE_COND)
>>>> + return false;
>>>> +
>>>> + tree cond = gimple_cond_lhs (gs);
>>>> +
>>>> + if (TREE_CODE (cond) != SSA_NAME)
>>>> + return false;
>>> space after if( too much vertical space in this function
>>> for my taste btw.
>> Will update this.
>>> For the forwarding to work we want a NE_EXPR or EQ_EXPR
>>> as gimple_cond_code and integer_one_p or integer_zero_p
>>> gimple_cond_rhs.
>> Right, checking those would be more safe. Since no issue found,
>during
>> bootstrap and regression tests, so I did not add these checking. I
>will
>> add this checking.
>Definitely want to verify that we're dealing with an equality test
>against 0/1.
>
>Jeff
next prev parent reply other threads:[~2019-05-30 6:40 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-05-21 13:45 Jiufu Guo
2019-05-22 12:38 ` Richard Biener
2019-05-23 12:06 ` Jiufu Guo
2019-05-23 12:11 ` Richard Biener
2019-05-23 14:40 ` Jiufu Guo
2019-05-24 12:45 ` Richard Biener
2019-05-24 14:52 ` Jiufu Guo
2019-05-28 14:07 ` [PATCH V2] " Jiufu Guo
2019-05-29 1:51 ` Jiufu Guo
2019-05-29 12:40 ` Richard Biener
2019-05-29 19:47 ` Jeff Law
2019-05-30 15:09 ` Jiufu Guo
2019-05-30 23:55 ` Jeff Law
2019-05-31 7:34 ` Richard Biener
2019-06-04 3:03 ` Jiufu Guo
2019-05-30 15:34 ` Jeff Law
2019-06-03 2:18 ` [PATCH V3] " Jiufu Guo
2019-06-04 5:30 ` [PATCH V4] " Jiufu Guo
2019-06-13 18:56 ` Jeff Law
2019-06-14 12:51 ` Jiufu Guo
2019-06-14 16:34 ` Jeff Law
2019-05-29 20:26 ` [PATCH] " Jeff Law
2019-05-30 6:57 ` Richard Biener
2019-05-30 6:58 ` Jiufu Guo
2019-05-30 14:59 ` Jeff Law
2019-05-30 15:03 ` Jeff Law
2019-05-29 20:22 ` Jeff Law
2019-05-30 6:40 ` Jiufu Guo
2019-05-30 6:44 ` Richard Biener
2019-05-30 20:17 ` Jeff Law
2019-05-31 7:30 ` Richard Biener
2019-05-31 15:28 ` Jeff Law
2019-06-04 5:19 ` Jiufu Guo
2019-06-04 7:07 ` Richard Biener
2019-06-07 0:05 ` Jeff Law
2019-05-29 20:18 ` Jeff Law
2019-05-30 6:41 ` Richard Biener [this message]
2019-05-29 20:12 ` Jeff Law
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=63105C16-2CE6-417D-A99A-7C13B45378B2@suse.de \
--to=rguenther@suse.de \
--cc=dberlin@dberlin.org \
--cc=gcc-patches@gcc.gnu.org \
--cc=guojiufu@linux.ibm.com \
--cc=jakub@redhat.com \
--cc=law@redhat.com \
--cc=segher@kernel.crashing.org \
--cc=wschmidt@linux.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).