All of lore.kernel.org
 help / color / mirror / Atom feed
From: Miquel Raynal <miquel.raynal@bootlin.com>
To: u-boot@lists.denx.de
Subject: [PATCH 06/17] fs/squashfs: sqfs_read_directory_table: fix memory leak
Date: Thu, 15 Oct 2020 18:38:51 +0200	[thread overview]
Message-ID: <20201015183851.01e8082f@xps13> (raw)
In-Reply-To: <6d4eecef-5d8f-37a2-a1df-1103aaab52d5@posteo.net>

Hi Richard,

Richard Genoud <richard.genoud@posteo.net> wrote on Thu, 15 Oct 2020
18:29:45 +0200:

> Hi Miquel !
> Thanks for your feedback.
> 
> Le 15/10/2020 ? 15:54, Miquel Raynal a ?crit?:
> > Hi Richard,
> > 
> > Richard Genoud <richard.genoud@posteo.net> wrote on Wed, 14 Oct 2020
> > 10:06:11 +0200:
> >   
> >> pos_list wasn't freed on every error
> >>
> >> Signed-off-by: Richard Genoud <richard.genoud@posteo.net>  
> > 
> > Same comment here (and probably after as well) as in patch 05/17, not
> > sure this is actually relevant for the community but I prefer this:
> > 
> > 	bar = malloc();
> > 	...
> > 	if (ret)
> > 		goto free_bar;
> > 
> > 	foo = malloc();
> > 	...
> > 	if (ret)
> > 		goto free foo;
> > 
> > 	...
> > 
> > 	foo:
> > 		kfree(foo);
> > 	bar:
> > 		kfree(bar);
> > 
> > than:
> > 
> > 	foo = NULL;
> > 	bar = NULL;
> > 
> > 	...
> > 	if (ret)
> > 		goto out;
> > 	...
> > 	if (ret)
> > 		goto out;
> > 	...
> > 		out:
> > 	if (ret)
> > 		kfree(...)  
> 
> I guess it's a coding habit.
> I personnaly prefer the later because I think it's less error-prone :
> When moving code aroung, we don't have to move the labels and rename
> the gotos.
> Ex:
> Let's say we have this code:
> 	bar = malloc();
> 	...
> 	if (ret)
> 		goto free_bar;
> 
> 	foo = malloc();
> 	...
> 	if (ret)
> 		goto free_foo;
> 	ret = init_somthing();
> 	if (ret)
> 		goto free_foo;
> 	ret = dummy()
> 	if (ret)
> 		goto free_foo;
> 
> 	...
> 
> 	foo:
> 		kfree(foo);
> 	bar:
> 		kfree(bar);
> 
> And, we want to move, for whatever reason, init_something() and dummy()
> before the foo allocation. We will have to change the code to:
> 
> 	bar = malloc();
> 	...
> 	if (ret)
> 		goto free_bar;
> 	ret = init_somthing();
> 	if (ret)
> 		goto free_bar; // not free_foo anymore !
> 	ret = dummy()
> 	if (ret)
> 		goto free_bar; // ditto
> 
> 	foo = malloc();
> 	...
> 	if (ret)
> 		goto free_foo;
> 	...
> 
> 	foo:
> 		kfree(foo);
> 	bar:
> 		kfree(bar);
> 
> Worse, if we have to exchange bar and foo allocation, we'll also have
> to exchange the deallocation of foo and bar and change all gotos beneath :
> 	foo = malloc();
> 	...
> 	if (ret)
> 		goto free_foo;
> 
> 	bar = malloc();
> 	...
> 	if (ret)
> 		goto free_bar;
> 
> 	ret = init_somthing();
> 	if (ret)
> 		goto free_foo; // not free_foo anymore
> 	ret = dummy()
> 	if (ret)
> 		goto free_foo; //ditto
> 
> 
> 	...
> 
> // oops ! we have to exchange that !
> 	foo:
> 		kfree(foo);
> 	bar:
> 		kfree(bar);
> 
> 
> That's why I prefer only one label and setting NULL.
> If I didn't convince you, I'll change it back to multiple labels :)

You are right it involves less changes when editing the code. But
on the other hand it is so often written like [my proposal], that it
almost becomes a coding style choice I guess. Anyway, I don't have a
strong opinion on this so I'll let you choose the best approach from
your point of view, unless you get other comments sharing my thoughts.

Thanks anyway for the cleanup :)

Cheers,
Miqu?l

  reply	other threads:[~2020-10-15 16:38 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-14  8:06 [PATCH 00/17] fs/squashfs: fix memory leaks and introduce exists() function Richard Genoud
2020-10-14  8:06 ` [PATCH 01/17] fs/squashfs: fix board hang-up when calling .exists() Richard Genoud
2020-10-14  8:06 ` [PATCH 02/17] fs/squashfs: sqfs_opendir: fix some memory leaks and dangling pointers Richard Genoud
2020-10-14  8:06 ` [PATCH 03/17] fs/squashfs: sqfs_opendir: simplify error handling Richard Genoud
2020-10-14  8:06 ` [PATCH 04/17] fs/squashfs: sqfs_closedir: fix memory leak Richard Genoud
2020-10-14  8:06 ` [PATCH 05/17] fs/squashfs: sqfs_split_path: fix memory leak and dangling pointers Richard Genoud
2020-10-15 13:49   ` Miquel Raynal
2020-10-14  8:06 ` [PATCH 06/17] fs/squashfs: sqfs_read_directory_table: fix memory leak Richard Genoud
2020-10-15 13:54   ` Miquel Raynal
2020-10-15 16:29     ` Richard Genoud
2020-10-15 16:38       ` Miquel Raynal [this message]
2020-10-16 12:31         ` Richard Genoud
2020-10-16 12:34           ` Miquel Raynal
2020-10-14  8:06 ` [PATCH 07/17] fs/squashfs: sqfs_search_dir: fix dangling pointer Richard Genoud
2020-10-14  8:06 ` [PATCH 08/17] fs/squashfs: sqfs_search_dir: fix memory leaks Richard Genoud
2020-10-14  8:06 ` [PATCH 09/17] fs/squashfs: sqfs_read_inode_table: fix dangling pointer Richard Genoud
2020-10-14  8:06 ` [PATCH 10/17] fs/squashfs: sqfs_concat_tokens: check if malloc succeeds Richard Genoud
2020-10-14  8:06 ` [PATCH 11/17] fs/squashfs: sqfs_size: fix dangling pointer dirs->entry Richard Genoud
2020-10-14  8:06 ` [PATCH 12/17] fs/squashfs: sqfs_size: remove useless sqfs_closedir() Richard Genoud
2020-10-14  8:06 ` [PATCH 13/17] fs/squashfs: sqfs_read: fix dangling pointer dirs->entry Richard Genoud
2020-10-14  8:06 ` [PATCH 14/17] fs/squashfs: sqfs_read: remove useless sqfs_closedir() Richard Genoud
2020-10-14  8:06 ` [PATCH 15/17] fs/squashfs: sqfs_read: fix memory leak Richard Genoud
2020-10-16 14:49   ` Richard Genoud
2020-10-14  8:06 ` [PATCH 16/17] fs/squashfs: sqfs_read: fix another " Richard Genoud
2020-10-14  8:06 ` [PATCH 17/17] fs/squashfs: implement exists() function Richard Genoud

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=20201015183851.01e8082f@xps13 \
    --to=miquel.raynal@bootlin.com \
    --cc=u-boot@lists.denx.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.