linux-kernel - Re: [RFC 11/15] PM / Hibernate: add chunk i/o support

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

Date:	Fri, 26 Mar 2010 21:02:28 +1100
From:	Nigel Cunningham <ncunningham@...a.org.au>
To:	Jiri Slaby <jirislaby@...il.com>
CC:	"Rafael J. Wysocki" <rjw@...k.pl>, pavel@....cz,
	linux-pm@...ts.linux-foundation.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC 11/15] PM / Hibernate: add chunk i/o support

Hi.

On 26/03/10 20:09, Jiri Slaby wrote:
> On 03/25/2010 11:38 PM, Rafael J. Wysocki wrote:
>>> +int sws_rw_buffer_init(int writing)
>>> +{
>>> +	BUG_ON(sws_writer_buffer || sws_writer_buffer_pos);
>>
>> Please don't do that.  Fail the operation instead.  You can also use WARN_ON
>> or WARN if you _really_ want the user to notice the failure.
>
> It's not a failure, it's a bug when we leak memory or forgot to
> read/write all data.
>
>> BUG_ON's like this are annoying like hell for testers who trigger them.
>
> I think BUG is appropriate here (the system or image is in an
> inconsitent state for the latter condition), but if you prefer the
> WARN-family here, I can switch it to that.
>
>>> +		if (writing) {
>>> +			ret = sws_io_ops->write_page(sws_writer_buffer, NULL);
>>> +			clear_page(sws_writer_buffer);
>>
>> Why do we need that clear_page()?
>
> Functionally for nothing, it was for my sakeness. Will remove.
>
>>> +int sws_rw_buffer_flush_page(int writing)
>>> +{
>>> +	int ret = 0;
>>> +	if (writing&&  sws_writer_buffer_pos)
>>> +		ret = sws_io_ops->write_page(sws_writer_buffer, NULL);
>>> +	sws_writer_buffer_pos = writing ? 0 : PAGE_SIZE;
>>> +	return ret;
>>> +}
>>
>> I'd split the above into two functions, one for writing and the other for
>> reading.
>>
>> Doing the same with sws_rw_buffer() (under a better name), for the sake of
>> clarity, also might make some sense, apparently.
>
> Do you mean adding hib*_buffer_read + hib*_buffer_write which would call
> static hib*_rw_buffer? sws_rw_buffer has much common code for R and W,
> so I would not make 2 functions from that.
>
> Nigel, you use _rw_ functions in toi, are there any pros opposing to _r_
> + _w_ (apart from exporting twice as symbols)?

I forget now why I used rw functions to begin with (it's been a long 
time!). I do know I've never worried about exporting twice as many symbols.

As I look at the code now, I think it makes more sense to split things 
up. This is especially true when I consider that a user with 4 cores has 
driven me to work on scalability, which will only diverge the code paths 
more.

Regards,

Nigel
--
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/