From: Daniel Mack <zonque@gmail.com>
To: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: linux-usb@vger.kernel.org, linux-omap@vger.kernel.org,
neumann@teufel.de, sergei.shtylyov@cogentembedded.com,
vinod.koul@intel.com, dan.j.williams@intel.com, balbi@ti.com,
gregkh@linuxfoundation.org
Subject: Re: [PATCH v3 4/5] dma: cppi41: only allocate descriptor memory once
Date: Tue, 01 Oct 2013 15:09:27 +0200 [thread overview]
Message-ID: <524AC987.5000301@gmail.com> (raw)
In-Reply-To: <20130926082604.GA24205@linutronix.de>
Hi Sebastian,
sorry for the long delay, I got distracted by other things.
On 26.09.2013 10:26, Sebastian Andrzej Siewior wrote:
> * Daniel Mack | 2013-09-22 16:50:03 [+0200]:
>
>> cdd->cd and cdd->descs_phys are allocated DESCS_AREAS times from
>> init_descs() and freed as often from purge_descs(). This leads to both
>> memory leaks and double-frees.
>>
>> Fix this by pulling the calls to dma_{alloc,free}_coherent() out of the
>> loops.
>>
>> While at it, remove the intermediate variable mem_decs (I guess it was
>> only there to make the code comply to the 80-chars CodingSytle rule).
>>
>> Signed-off-by: Daniel Mack <zonque@gmail.com>
>
> Please don't merge the memory descriptors. The idea was initially to
> allocate multiple small descriptors instead one big. The descrriptor
> turned out to be enough so it looks like the way it looks.
> If you want to clean this up, please either remove the for loop and
> allocate only one memory area or please prepare for multiple descripors
> (I think two is the upper limit).
Well, I didn't merge the descriptors. Look again at my changes please.
A simplified version of the code as it stands is:
for (i = 0; i < DESCS_AREAS; i++)
cdd->cd = dma_alloc_coherent(dev, ..., &cdd->descs_phys, GFP_KERNEL);
for (i = 0; i < DESCS_AREAS; i++)
dma_free_coherent(dev, mem_decs, cdd->cd, cdd->descs_phys);
So you're effectively allocating and freeing the same pointer
DESCS_AREAS times, which is certainly not what you wanted.
And this just doesn't hit you because DESCS_AREAS is always 1:
BUILD_BUG_ON(DESCS_AREAS != 1);
So, after all, my patch doesn't really change any of the runtime
behaviour. Consider it a cosmetic cleanup if you wish :)
Thanks,
Daniel
next prev parent reply other threads:[~2013-10-01 13:09 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-09-22 14:49 [PATCH v3 0/5] dma: cppi41: some trivial fixes and support for suspend/resume Daniel Mack
[not found] ` <1379861404-8250-1-git-send-email-zonque-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2013-09-22 14:50 ` [PATCH v3 1/5] dma: cppi41: pass around device instead of platform_device Daniel Mack
2013-09-23 4:16 ` Vinod Koul
2013-09-22 14:50 ` [PATCH v3 2/5] dma: cppi41: s/deinit_cpii41/deinit_cppi41/ Daniel Mack
2013-09-23 4:16 ` Vinod Koul
2013-09-22 14:50 ` [PATCH v3 3/5] dma: cppi41: add shortcut to &pdev->dev in cppi41_dma_probe() Daniel Mack
[not found] ` <1379861404-8250-4-git-send-email-zonque-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2013-09-23 4:17 ` Vinod Koul
2013-09-22 14:50 ` [PATCH v3 4/5] dma: cppi41: only allocate descriptor memory once Daniel Mack
2013-09-23 4:17 ` Vinod Koul
[not found] ` <20130923041754.GZ17188-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2013-09-23 14:51 ` Sebastian Andrzej Siewior
2013-09-23 14:36 ` Vinod Koul
[not found] ` <1379861404-8250-5-git-send-email-zonque-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2013-09-26 8:26 ` Sebastian Andrzej Siewior
2013-10-01 13:09 ` Daniel Mack [this message]
[not found] ` <524AC987.5000301-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2013-10-01 16:22 ` Sebastian Andrzej Siewior
2013-10-01 16:57 ` Daniel Mack
2013-10-02 7:13 ` Sebastian Andrzej Siewior
2013-09-22 14:50 ` [PATCH v3 5/5] dma: cppi41: add support for suspend and resume Daniel Mack
[not found] ` <1379861404-8250-6-git-send-email-zonque-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2013-09-23 4:09 ` Vinod Koul
2013-09-23 5:53 ` Daniel Mack
2013-09-23 10:00 ` Vinod Koul
2013-09-23 10:01 ` Vinod Koul
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=524AC987.5000301@gmail.com \
--to=zonque@gmail.com \
--cc=balbi@ti.com \
--cc=bigeasy@linutronix.de \
--cc=dan.j.williams@intel.com \
--cc=gregkh@linuxfoundation.org \
--cc=linux-omap@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=neumann@teufel.de \
--cc=sergei.shtylyov@cogentembedded.com \
--cc=vinod.koul@intel.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 NNTP newsgroup(s).