Re: [RFC][PATCH 6/7] MTD: UBI: Implement checkpointing support

[Shmulik Ladkani <shmulik.ladkani@gmail.com>]

On Tue, 14 Feb 2012 21:06:45 +0100 Richard Weinberger <rw@linutronix.de> wrote:
> Implements UBI checkpointing support.
> It reduces the attaching time from O(N) to O(1).
> Checkpoints are written on demand and upon changes of the volume layout.
> If the recovery from a checkpoint fails we fall back to scanning mode.

Partially reviewed the feature. Great work.
There's some tiny styling/coding issues, will send references if you'd
like.

I'll comment on the feature itself later on.

Meanwhile, there's a potential memleak/crash you might wanna fix.

> +/* Reads the checkpoint data from it's PEBs */
> +struct ubi_scan_info *ubi_read_checkpoint(struct ubi_device *ubi, int cb_sb_pnum)
> +{
> +	struct ubi_cp_sb *cpsb;
> +	struct ubi_vid_hdr *vh;
> +	int ret, i, nblocks;
> +	char *cp_raw;
> +	size_t cp_size;
> +	__be32 data_crc;
> +	unsigned long long sqnum = 0;
> +	struct ubi_scan_info *si = NULL;
> +
> +	cpsb = kmalloc(sizeof(*cpsb), GFP_KERNEL);
> +	if (!cpsb) {
> +		si = ERR_PTR(-ENOMEM);
> +		goto out;
> +	}
> +
> +	ret = ubi_io_read(ubi, cpsb, cb_sb_pnum, ubi->leb_start, sizeof(*cpsb));
> +	if (ret) {
> +		ubi_err("Unable to read checkpoint super block");
> +		si = ERR_PTR(ret);
> +		goto out;

s/goto out/goto free_sb/
(otherwise 'cpsb' not freed)

> +	/* cp_raw will contain the whole checkpoint */
> +	cp_raw = vzalloc(cp_size);

  ...

> +
> +	cpsb = (struct ubi_cp_sb *)cp_raw;

'cpsb' is overwritten, but formerly kmalloced (at the beginning of
ubi_read_checkpoint).
Should free 'cpsb' prior assignment, or alternatively use different
variable then 'cpsb'.

  ...

> +
> +free_vhdr:
> +	ubi_free_vid_hdr(ubi, vh);
> +free_raw:
> +	vfree(cp_raw);
> +free_sb:
> +	kfree(cpsb);

Freeing 'cp_raw' and 'cpsb', but in the normal flow, they point to the
same thing.

Regards,
Shmulik