linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ondrej Zary <linux@rainbow-software.org>
To: "Benny Sjöstrand" <benny@hostmobility.com>
Cc: Jesper Juhl <jj@chaosbits.net>,
	linux-kernel@vger.kernel.org, Jaroslav Kysela <perex@perex.cz>,
	Takashi Iwai <tiwai@suse.de>,
	alsa-devel@alsa-project.org
Subject: Re: [PATCH] cs46xx memory management fixes for cs46xx_dsp_spos_create() - make sure we free and don't do pointless work.
Date: Sat, 30 Oct 2010 14:06:07 +0200	[thread overview]
Message-ID: <201010301406.12961.linux@rainbow-software.org> (raw)
In-Reply-To: <4CCB4851.7080701@hostmobility.com>

On Saturday 30 October 2010 00:18:57 Benny Sjöstrand wrote:
> Hello again!
>
> Just doing a reply-all.
> It's has been many years since I did anything to the cs46xx driver, so
> I'm wondering
> if there's anyone out there still using a cs46xx sound card?
> I think the changes look's correct, but I can't test it, I do not have a
> cs46xx hardware anymore.

Sure, there are many of these cards "out there". I just got a "new" Terratec 
DMX Xfire 1024 (CS4624). I'll test the patch.

>
> /Benny
>
> Jesper Juhl wrote:
> > Hi,
> >
> > When reading through sound/pci/cs46xx/dsp_spos.c I noticed a couple of
> > things in cs46xx_dsp_spos_create().
> >
> > It seems to me that we don't always free the various memory buffers we
> > allocate and we also do some work (structure member assignment) early,
> > that is completely pointless if some of the memory allocations fail and
> > we end up just aborting the whole thing.
> >
> > I don't have hardware to test, so the patch below is compile tested only,
> > but it makes the following changes:
> >
> > - Make sure we always free all allocated memory on failures.
> > - Don't do pointless work assigning to structure members before we know
> >   all memory allocations, that may abort progress, have completed
> >   successfully.
> > - Remove some trailing whitespace.
> >
> > If it looks ok, please merge, otherwise I'd be interested in knowing
> > what's wrong so I can improve it.
> >
> >
> > Signed-off-by: Jesper Juhl <jj@chaosbits.net>
> > ---
> >  dsp_spos.c |   33 +++++++++++----------------------
> >  1 file changed, 11 insertions(+), 22 deletions(-)
> >
> > diff --git a/sound/pci/cs46xx/dsp_spos.c b/sound/pci/cs46xx/dsp_spos.c
> > index 3e5ca8f..e377287 100644
> > --- a/sound/pci/cs46xx/dsp_spos.c
> > +++ b/sound/pci/cs46xx/dsp_spos.c
> > @@ -225,39 +225,25 @@ struct dsp_spos_instance *cs46xx_dsp_spos_create
> > (struct snd_cs46xx * chip) {
> >  	struct dsp_spos_instance * ins = kzalloc(sizeof(struct
> > dsp_spos_instance), GFP_KERNEL);
> >
> > -	if (ins == NULL)
> > +	if (ins == NULL)
> >  		return NULL;
> >
> >  	/* better to use vmalloc for this big table */
> > -	ins->symbol_table.nsymbols = 0;
> >  	ins->symbol_table.symbols = vmalloc(sizeof(struct dsp_symbol_entry) *
> >  					    DSP_MAX_SYMBOLS);
> > -	ins->symbol_table.highest_frag_index = 0;
> > -
> > -	if (ins->symbol_table.symbols == NULL) {
> > +	ins->code.data = kmalloc(DSP_CODE_BYTE_SIZE, GFP_KERNEL);
> > +	ins->modules = kmalloc(sizeof(struct dsp_module_desc) *
> > DSP_MAX_MODULES, GFP_KERNEL); +	if (!ins->symbol_table.symbols ||
> > !ins->code.data || !ins->modules) { cs46xx_dsp_spos_destroy(chip);
> >  		goto error;
> >  	}
> > -
> > +	ins->symbol_table.nsymbols = 0;
> > +	ins->symbol_table.highest_frag_index = 0;
> >  	ins->code.offset = 0;
> >  	ins->code.size = 0;
> > -	ins->code.data = kmalloc(DSP_CODE_BYTE_SIZE, GFP_KERNEL);
> > -
> > -	if (ins->code.data == NULL) {
> > -		cs46xx_dsp_spos_destroy(chip);
> > -		goto error;
> > -	}
> > -
> >  	ins->nscb = 0;
> >  	ins->ntask = 0;
> > -
> >  	ins->nmodules = 0;
> > -	ins->modules = kmalloc(sizeof(struct dsp_module_desc) *
> > DSP_MAX_MODULES, GFP_KERNEL); -
> > -	if (ins->modules == NULL) {
> > -		cs46xx_dsp_spos_destroy(chip);
> > -		goto error;
> > -	}
> >
> >  	/* default SPDIF input sample rate
> >  	   to 48000 khz */
> > @@ -271,8 +257,8 @@ struct dsp_spos_instance *cs46xx_dsp_spos_create
> > (struct snd_cs46xx * chip)
> >
> >  	/* set left and right validity bits and
> >  	   default channel status */
> > -	ins->spdif_csuv_default =
> > -		ins->spdif_csuv_stream =
> > +	ins->spdif_csuv_default =
> > +		ins->spdif_csuv_stream =
> >  	 /* byte 0 */  ((unsigned int)_wrap_all_bits( 
> > (SNDRV_PCM_DEFAULT_CON_SPDIF        & 0xff)) << 24) | /* byte 1 */ 
> > ((unsigned int)_wrap_all_bits( ((SNDRV_PCM_DEFAULT_CON_SPDIF >> 8) &
> > 0xff)) << 16) | /* byte 3 */   (unsigned int)_wrap_all_bits( 
> > (SNDRV_PCM_DEFAULT_CON_SPDIF >> 24) & 0xff) | @@ -281,6 +267,9 @@ struct
> > dsp_spos_instance *cs46xx_dsp_spos_create (struct snd_cs46xx * chip)
> > return ins;
> >
> >  error:
> > +	kfree(ins->modules);
> > +	kfree(ins->code.data);
> > +	vfree(ins->symbol_table.symbols);
> >  	kfree(ins);
> >  	return NULL;
> >  }
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/



-- 
Ondrej Zary

  reply	other threads:[~2010-10-30 12:06 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-10-29 20:54 [PATCH] cs46xx memory management fixes for cs46xx_dsp_spos_create() - make sure we free and don't do pointless work Jesper Juhl
2010-10-29 22:18 ` Benny Sjöstrand
2010-10-30 12:06   ` Ondrej Zary [this message]
2010-10-30 19:44     ` Jesper Juhl
2010-10-31 17:29       ` Ondrej Zary
2010-11-01  9:38         ` Takashi Iwai
2010-11-01 18:23           ` Jesper Juhl

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=201010301406.12961.linux@rainbow-software.org \
    --to=linux@rainbow-software.org \
    --cc=alsa-devel@alsa-project.org \
    --cc=benny@hostmobility.com \
    --cc=jj@chaosbits.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=perex@perex.cz \
    --cc=tiwai@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).