On Fri, Jul 31, 2015 at 12:12:45PM +0300, Amarnath Valluri wrote:
> * libparted/fs/r/hfs/hfs.c(hfsplus_resize): Don't leak embedded_geom
> when volume resizing failed.
> * libparted/fs/fat/fat.c(fat_create): Don't leake fs in-case of
> fat_write_clustor failuer.
> * libparted/arch/linux.c(_kernel_get_partition_start_and_length): Don't leak
> dev_fd file descriptor.
>
> Signed-off-by: Amarnath Valluri <[email protected]>
> ---
> libparted/arch/linux.c | 1 +
> libparted/fs/r/fat/fat.c | 2 +-
> libparted/fs/r/hfs/hfs.c | 1 +
> 3 files changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/libparted/arch/linux.c b/libparted/arch/linux.c
> index 27bbb80..349442b 100644
> --- a/libparted/arch/linux.c
> +++ b/libparted/arch/linux.c
> @@ -2665,6 +2665,7 @@ _kernel_get_partition_start_and_length(PedPartition
> const *part,
> int dev_fd = open (dev_name, O_RDONLY);
> if (dev_fd != -1 && ioctl (dev_fd, HDIO_GETGEO, &geom)) {
> *start = geom.start;
> + close (dev_fd);
> ok = true;
> } else {
> if (dev_fd != -1)
> diff --git a/libparted/fs/r/fat/fat.c b/libparted/fs/r/fat/fat.c
> index 5aa72d8..4ecf5c5 100644
> --- a/libparted/fs/r/fat/fat.c
> +++ b/libparted/fs/r/fat/fat.c
> @@ -305,7 +305,7 @@ fat_create (PedGeometry* geom, FatType fat_type,
> PedTimer* timer)
> memset (fs_info->buffer, 0, fs_info->cluster_size);
> if (!fat_write_cluster (fs, fs_info->buffer,
> fs_info->root_cluster))
> - return 0;
> + goto error_free_buffers;
> }
The ones above here look good to me.
>
> fs_info->serial_number = generate_random_uint32 ();
> diff --git a/libparted/fs/r/hfs/hfs.c b/libparted/fs/r/hfs/hfs.c
> index a1c37cc..61edcbb 100644
> --- a/libparted/fs/r/hfs/hfs.c
> +++ b/libparted/fs/r/hfs/hfs.c
> @@ -985,6 +985,7 @@ hfsplus_resize (PedFileSystem* fs, PedGeometry* geom,
> PedTimer* timer)
> /* Resize the HFS+ volume */
> if (!hfsplus_volume_resize (fs, embedded_geom, timer_plus)) {
> if (timer_plus != timer) ped_timer_destroy_nested (timer_plus);
> + ped_geometry_destroy (embedded_geom);
> ped_exception_throw (
> PED_EXCEPTION_ERROR,
> PED_EXCEPTION_CANCEL,
I think this needs to be wrapped in a priv_data->wrapper check,
otherwise it'll end up destroying the geom that's passed in from the
caller.
--
Brian C. Lane | Anaconda Team | IRC: bcl #anaconda | Port Orchard, WA (PST8PDT)
