From: Wei Yang <richard.weiyang@gmail.com>
To: NeilBrown <neilb@suse.de>
Cc: Wei Yang <richard.weiyang@gmail.com>,
kuba@kernel.org, gregkh@linuxfoundation.org,
mojha@codeaurora.org, jkosina@suse.cz,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH] hashtable: remove a redundant check in hash_for_each_xxx()
Date: Thu, 7 Oct 2021 00:30:58 +0000 [thread overview]
Message-ID: <20211007003058.uj35ekwibbrxqzku@master> (raw)
In-Reply-To: <163355497171.31063.8329134032738647570@noble.neil.brown.name>
On Thu, Oct 07, 2021 at 08:16:11AM +1100, NeilBrown wrote:
>On Thu, 07 Oct 2021, Wei Yang wrote:
>> The three hash_for_each_xxx() helper iterate the hash table with help
>> of hlist_for_each_entry_xxx(), which breaks the loop only when obj is
>> NULL.
>>
>> This means the check during each iteration is redundant. This patch
>> removes it.
>>
>> Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
>> ---
>> include/linux/hashtable.h | 9 +++------
>> 1 file changed, 3 insertions(+), 6 deletions(-)
>>
>> diff --git a/include/linux/hashtable.h b/include/linux/hashtable.h
>> index f6c666730b8c..a15719ed303f 100644
>> --- a/include/linux/hashtable.h
>> +++ b/include/linux/hashtable.h
>> @@ -124,8 +124,7 @@ static inline void hash_del_rcu(struct hlist_node *node)
>> * @member: the name of the hlist_node within the struct
>> */
>> #define hash_for_each(name, bkt, obj, member) \
>> - for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
>> - (bkt)++)\
>> + for ((bkt) = 0, obj = NULL; (bkt) < HASH_SIZE(name); (bkt)++) \
>> hlist_for_each_entry(obj, &name[bkt], member)
>
>I think you are missing an important property of this code.
>What we have here is a new loop command (hash_for_each()) that is
>constructed from 2 nested loops. This sort of construct is in general
>difficult to use because in C it is common to use "break" to exit a loop
>early. 'break' cannot exit two levels of loop though. So if you aren't
>careful, doing something like
>
> hash_for_each() {
> do something
> if (some test)
> break;
> }
>
>might not do what you expect. The 'break' will exit the inner loop, but
>not the outer loop. That could easily lead to buggy code.
>
>But this macro *is* careful. If the loop body *does* use break, then
>the inner loop will abort but 'obj' will still be non-NULL. The test
>for NULL in the outer loop causes the outer loop to abort too - as the
>programmer probably expected.
>
Thanks for pointing out. I missed this case.
>So by removing the 'obj == NULL' test, you would cause any usage which
>breaks out of the loop to now be incorrect.
>
>I recommend that instead of this patch, you provide a patch which
>improves the documentation to make this clear. e.g.
>
> Note: it is safe to 'break' out of this loop even though it is a two
> nested loops. The 'obj == NULL' test ensures that when the inner loop
> is broken, the outer loop will break too.
>
Here is a draft patch based on you comment:
diff --git a/include/linux/hashtable.h b/include/linux/hashtable.h
index f6c666730b8c..2ff4cb5e6a22 100644
--- a/include/linux/hashtable.h
+++ b/include/linux/hashtable.h
@@ -116,6 +116,13 @@ static inline void hash_del_rcu(struct hlist_node *node)
hlist_del_init_rcu(node);
}
+/**
+ * Note: the following three hash_for_each[_xxx] helpers introduce a new loop
+ * command that is constructed from 2 nested loops. It is safe to 'break' out
+ * of this loop even though it is a two nested loops. The 'obj == NULL' test
+ * ensures that when the inner loop is broken, the outer loop will break too.
+ */
+
/**
* hash_for_each - iterate over a hashtable
* @name: hashtable to iterate
If you feel good, I would like to add
Sugguested-by: NeilBrown <neilb@suse.de>
>Thanks,
>NeilBrown
>
>
>>
>> /**
>> @@ -136,8 +135,7 @@ static inline void hash_del_rcu(struct hlist_node *node)
>> * @member: the name of the hlist_node within the struct
>> */
>> #define hash_for_each_rcu(name, bkt, obj, member) \
>> - for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
>> - (bkt)++)\
>> + for ((bkt) = 0, obj = NULL; (bkt) < HASH_SIZE(name); (bkt)++) \
>> hlist_for_each_entry_rcu(obj, &name[bkt], member)
>>
>> /**
>> @@ -150,8 +148,7 @@ static inline void hash_del_rcu(struct hlist_node *node)
>> * @member: the name of the hlist_node within the struct
>> */
>> #define hash_for_each_safe(name, bkt, tmp, obj, member) \
>> - for ((bkt) = 0, obj = NULL; obj == NULL && (bkt) < HASH_SIZE(name);\
>> - (bkt)++)\
>> + for ((bkt) = 0, obj = NULL; (bkt) < HASH_SIZE(name); (bkt)++) \
>> hlist_for_each_entry_safe(obj, tmp, &name[bkt], member)
>>
>> /**
>> --
>> 2.23.0
>>
>>
--
Wei Yang
Help you, Help me
next prev parent reply other threads:[~2021-10-07 0:31 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-06 15:21 [PATCH] hashtable: remove a redundant check in hash_for_each_xxx() Wei Yang
2021-10-06 15:29 ` Greg KH
2021-10-06 21:16 ` NeilBrown
2021-10-07 0:30 ` Wei Yang [this message]
2021-10-07 0:50 ` NeilBrown
2021-10-07 23:40 ` Wei Yang
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=20211007003058.uj35ekwibbrxqzku@master \
--to=richard.weiyang@gmail.com \
--cc=gregkh@linuxfoundation.org \
--cc=jkosina@suse.cz \
--cc=kuba@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mojha@codeaurora.org \
--cc=neilb@suse.de \
/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 NNTP newsgroup(s).