linux-kernel - Re: [PATCH] Introduce buflock, a one-to-many circular buffer mechanism (rev2)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <4C20C098.1010200@euromail.se>
Date:	Tue, 22 Jun 2010 15:54:32 +0200
From:	Henrik Rydberg <rydberg@...omail.se>
To:	Andrew Morton <akpm@...ux-foundation.org>
CC:	Dmitry Torokhov <dmitry.torokhov@...il.com>,
	linux-input@...r.kernel.org, linux-kernel@...r.kernel.org,
	Oleg Nesterov <oleg@...hat.com>, Jiri Kosina <jkosina@...e.cz>,
	Jonathan Cameron <jic23@....ac.uk>
Subject: Re: [PATCH] Introduce buflock, a one-to-many circular buffer mechanism
 (rev2)

Andrew Morton wrote:
[...]
> I don't understand why this has "lock" in its name.
> 
> The API itself is a mixture of "bufwrite_foo" and "bufread_foo".
> 
> It's all a bit chaotic.  I'd suggest picking a sane name for the whole
> subsytem - perhaps "mrbuf" for "multi reader buffer"?  Then
> consistently name all interface functions as "mrbuf_foo". 
> mrbuf.h, mrbuf_write_lock(), etc.

Point taken.

>> +static __always_inline bool __must_check bufread_retry(struct buflock_reader *br, const struct buflock_writer *bw)
>> +{
>> +	smp_rmb();
>> +	if (unlikely(((br->tail - br->last) & bw->page) < bw->next - br->last))
>> +		return true;
>> +	++br->tail;
>> +	if (unlikely(br->head - br->tail > bw->page))
>> +		br->tail = br->head;
>> +	return false;
>> +}
> 
> This looks too large to be inlined.
> 
> What's the __always_inline for?  Was gcc uninlining this within
> separate compilation units?

As you say, the function is large, and I am uncertain about the rules regarding
compiler reordering across general function calls. Starting a general function
with a memory barrier feels weird. Perhaps the function should be split?

Thanks,
Henrik

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/