linux-kernel - Re: [PATCH 10/22 take 3] UBI: EBA unit

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 15 Mar 2007 11:07:03 -0800
From:	Andrew Morton <akpm@...ux-foundation.org>
To:	Artem Bityutskiy <dedekind@...radead.org>
Cc:	linux-kernel@...r.kernel.org, haver@...t.ibm.com,
	hch@...radead.org, dwmw2@...radead.org, jwboyer@...ux.vnet.ibm.com,
	dedekind@...radead.org
Subject: Re: [PATCH 10/22 take 3] UBI: EBA unit


There's way too much code here to expect it to get decently reviewed, alas.

> On Wed, 14 Mar 2007 17:20:24 +0200 Artem Bityutskiy <dedekind@...radead.org> wrote:
>
> ...
>
> +/**
> + * leb_get_ver - get logical eraseblock version.
> + *
> + * @ubi: the UBI device description object
> + * @vol_id: the volume ID
> + * @lnum: the logical eraseblock number
> + *
> + * The logical eraseblock has to be locked. Note, all this leb_ver stuff is
> + * obsolete and will be removed eventually. FIXME: to be removed together with
> + * leb_ver support.
> + */
> +static inline int leb_get_ver(struct ubi_info *ubi, int vol_id, int lnum)
> +{
> +	int idx, leb_ver;
> +
> +	idx = vol_id2idx(ubi, vol_id);
> +
> +	spin_lock(&ubi->eba.eba_tbl_lock);
> +	ubi_assert(ubi->eba.eba_tbl[idx].recs);
> +	leb_ver = ubi->eba.eba_tbl[idx].recs[lnum].leb_ver;
> +	spin_unlock(&ubi->eba.eba_tbl_lock);
> +
> +	return leb_ver;
> +}

I very much doubt that the locking in this function (and in the similar
ones here) does anything useful.

> +static unsigned long long next_sqnum(struct ubi_info *ubi)
> +{
> +	unsigned long long sqnum;
> +
> +	spin_lock(&ubi->eba.eba_tbl_lock);
> +	sqnum = ubi->eba.global_sq_cnt++;
> +	spin_unlock(&ubi->eba.eba_tbl_lock);
> +
> +	return sqnum;
> +}

That one makes sense,

> +static inline void leb_map(struct ubi_info *ubi, int vol_id, int lnum, int pnum)
> +{
> +	int idx;
> +
> +	idx = vol_id2idx(ubi, vol_id);
> +	spin_lock(&ubi->eba.eba_tbl_lock);
> +	ubi_assert(ubi->eba.eba_tbl[idx].recs);
> +	ubi_assert(ubi->eba.eba_tbl[idx].recs[lnum].pnum < 0);
> +	ubi->eba.eba_tbl[idx].recs[lnum].pnum = pnum;
> +	spin_unlock(&ubi->eba.eba_tbl_lock);
> +}

I doubt if that one does.

> +/**
> + * leb_unmap - un-map a logical eraseblock.
> + *
> + * @ubi: the UBI device description object
> + * @vol_id: the volume ID
> + * @lnum: the logical eraseblock number to unmap
> + *
> + * This function un-maps a logical eraseblock and increases its version. The
> + * logical eraseblock has to be locked.
> + */
> +static inline void leb_unmap(struct ubi_info *ubi, int vol_id, int lnum)

The patch is full of nutty inlining.

Suggestion: just remove all of it.  Then reintroduce inlining in only
those places where a benefit is demonstrable.  Reduced code size according to
/bin/size would be a suitable metric.

> +static inline int leb2peb(struct ubi_info *ubi, int vol_id, int lnum)
> +{
> +	int idx, pnum;
> +
> +	idx = vol_id2idx(ubi, vol_id);
> +
> +	spin_lock(&ubi->eba.eba_tbl_lock);
> +	ubi_assert(ubi->eba.eba_tbl[idx].recs);
> +	pnum = ubi->eba.eba_tbl[idx].recs[lnum].pnum;
> +	spin_unlock(&ubi->eba.eba_tbl_lock);
> +
> +	return pnum;
> +}

Again, the locking seems pointless.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/