Re: [PATCH 1/2] ubifs: Fix potential memory leaks while iterating entries
On Mon, Jun 1, 2020 at 3:46 PM Zhihao Cheng wrote: > > 在 2020/6/1 20:00, Markus Elfring 写道: > >> Fix some potential memory leaks in error handling branches while > >> iterating xattr entries. > > Such information is useful. > > > > > >> For example, function ubifs_tnc_remove_ino() > >> forgets to free pxent if it exists. Similar problems also exist in > >> ubifs_purge_xattrs(), ubifs_add_orphan() and ubifs_jnl_write_inode(). > > Can an other wording variant be a bit nicer? > Thanks for reminding, I will improve this description. The description is fine. No need to resend. Patch will be queued after the merge window. Thanks for fixing these issues! -- Thanks, //richard
Re: [PATCH 1/2] ubifs: Fix potential memory leaks while iterating entries
在 2020/6/1 20:00, Markus Elfring 写道: Fix some potential memory leaks in error handling branches while iterating xattr entries. Such information is useful. For example, function ubifs_tnc_remove_ino() forgets to free pxent if it exists. Similar problems also exist in ubifs_purge_xattrs(), ubifs_add_orphan() and ubifs_jnl_write_inode(). Can an other wording variant be a bit nicer? Thanks for reminding, I will improve this description. I suggest to avoid the specification of duplicate function calls (also for the desired exception handling). Will it be helpful to add a few jump targets? https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/coding-style.rst?id=3d77e6a8804abcc0504c904bd6e5cdf3a5cf8162#n455 I've thought about using "goto err_tag_2" in kill_xattrs code block to release prev xent, but later it needs to jump to 'out_release tag‘ for releasing previously requested memory, which can clutter the code. It seems that two consecutive 'goto tags' will make the code less readable. Regards, Markus .
Re: [PATCH 1/2] ubifs: Fix potential memory leaks while iterating entries
> Fix some potential memory leaks in error handling branches while > iterating xattr entries. Such information is useful. > For example, function ubifs_tnc_remove_ino() > forgets to free pxent if it exists. Similar problems also exist in > ubifs_purge_xattrs(), ubifs_add_orphan() and ubifs_jnl_write_inode(). Can an other wording variant be a bit nicer? I suggest to avoid the specification of duplicate function calls (also for the desired exception handling). Will it be helpful to add a few jump targets? https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/coding-style.rst?id=3d77e6a8804abcc0504c904bd6e5cdf3a5cf8162#n455 Regards, Markus
