linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Richard Weinberger <richard@nod.at>
To: linux-mtd@lists.infradead.org
Cc: linux-kernel@vger.kernel.org, Richard Weinberger <richard@nod.at>
Subject: [PATCH 6/8] ubi: fastmap: Scan empty space if fastmap is preseeded
Date: Mon, 14 May 2018 13:24:20 +0200	[thread overview]
Message-ID: <20180514112422.23988-7-richard@nod.at> (raw)
In-Reply-To: <20180514112422.23988-1-richard@nod.at>

The creation tool does not know the real size of the MTD, therefore the
preseeded fastmap references only used PEBs.
Free PEBs need to be discovered during the initial attach of the
preseeded fastmap.

Signed-off-by: Richard Weinberger <richard@nod.at>
---
 drivers/mtd/ubi/fastmap.c | 109 +++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 103 insertions(+), 6 deletions(-)

diff --git a/drivers/mtd/ubi/fastmap.c b/drivers/mtd/ubi/fastmap.c
index 0d1274b69274..5ec14fd70c13 100644
--- a/drivers/mtd/ubi/fastmap.c
+++ b/drivers/mtd/ubi/fastmap.c
@@ -400,6 +400,94 @@ static void unmap_peb(struct ubi_attach_info *ai, int pnum)
 		}
 	}
 }
+/**
+ * scan_empty_space - scan empty space.
+ *
+ * @ubi: UBI device object
+ * @ai: attach info object
+ * @start: PEB number where empty space is expected to start
+ *
+ * Is is the fastmap preseeded, it references only used PEBs, the creation
+ * does not know the real MTD size. Therefore many PEBs are 0xff and unknown
+ * to the fastmap. Scan for this PEBs during attach and make them known.
+ * These PEBs are only allowed to be 0xff or have a valid EC header.
+ * EC headers are allowed because the initial scan+erase operation could be
+ * interrupted by a power cycle.
+ */
+static int scan_empty_space(struct ubi_device *ubi, struct ubi_attach_info *ai,
+			    int start)
+{
+	int pnum, err, scrub, empty, image_seq;
+	unsigned long long ec;
+	struct ubi_ec_hdr *ech = NULL;
+	struct ubi_vid_io_buf *vb = NULL;
+	struct ubi_vid_hdr *vh;
+
+	err = -ENOMEM;
+
+	ech = kzalloc(ubi->ec_hdr_alsize, GFP_KERNEL);
+	if (!ech)
+		goto out;
+
+	vb = ubi_alloc_vid_buf(ubi, GFP_KERNEL);
+	if (!vb)
+		goto out;
+
+	vh = ubi_get_vid_hdr(vb);
+
+	for (pnum = start; pnum < ubi->peb_count; pnum++) {
+		if (ubi_io_is_bad(ubi, pnum))
+			continue;
+
+		scrub = empty = ec = 0;
+		err = ubi_io_read_ec_hdr(ubi, pnum, ech, 0);
+		switch (err) {
+			case UBI_IO_FF_BITFLIPS:
+				scrub = 1;
+				/* fall through */
+			case UBI_IO_FF:
+				empty = 1;
+				break;
+			case UBI_IO_BITFLIPS:
+				scrub = 1;
+				/* fall through */
+			case 0:
+				ec = be64_to_cpu(ech->ec);
+				image_seq = be32_to_cpu(ech->image_seq);
+
+				if (image_seq && (image_seq != ubi->image_seq)) {
+					ubi_err(ubi, "bad image seq: 0x%x, expected: 0x%x",
+						image_seq, ubi->image_seq);
+					err = UBI_BAD_FASTMAP;
+					goto out;
+				}
+				break;
+			default:
+				ubi_err(ubi, "Unable to read EC header in empty space: %i",
+					err);
+				err = UBI_BAD_FASTMAP;
+				goto out;
+		}
+
+		err = ubi_io_read_vid_hdr(ubi, pnum, vb, 0);
+		if (err != UBI_IO_FF && err != UBI_IO_FF_BITFLIPS) {
+			ubi_err(ubi, "Unexpected data in empty space: %i", err);
+			err = UBI_BAD_FASTMAP;
+			goto out;
+		}
+
+		if (empty)
+			add_aeb(ai, &ai->erase, pnum, ec, scrub);
+		else
+			add_aeb(ai, &ai->free, pnum, ec, scrub);
+	}
+
+	err = 0;
+out:
+	kfree(ech);
+	ubi_free_vid_buf(vb);
+	return err;
+}
 
 /**
  * scan_pool - scans a pool for changed (no longer empty PEBs).
@@ -784,13 +872,22 @@ static int ubi_attach_fastmap(struct ubi_device *ubi,
 		}
 	}
 
-	ret = scan_pool(ubi, ai, fmpl->pebs, pool_size, &max_sqnum, &free);
-	if (ret)
-		goto fail;
+	if (!(fm->flags & UBI_FM_SB_PRESEEDED_FLG)) {
+		ret = scan_pool(ubi, ai, fmpl->pebs, pool_size, &max_sqnum, &free);
+		if (ret)
+			goto fail;
 
-	ret = scan_pool(ubi, ai, fmpl_wl->pebs, wl_pool_size, &max_sqnum, &free);
-	if (ret)
-		goto fail;
+		ret = scan_pool(ubi, ai, fmpl_wl->pebs, wl_pool_size, &max_sqnum, &free);
+		if (ret)
+			goto fail;
+	}
+
+	if (fm->flags & UBI_FM_SB_PRESEEDED_FLG) {
+		int empty_start = fix_pnum(ai, be32_to_cpu(fmhdr->used_peb_count) + fm->used_blocks);
+		ret = scan_empty_space(ubi, ai, empty_start);
+		if (ret)
+			goto fail;
+	}
 
 	if (max_sqnum > ai->max_sqnum)
 		ai->max_sqnum = max_sqnum;
-- 
2.13.6

  parent reply	other threads:[~2018-05-14 11:25 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-05-14 11:24 [PATCH 0/8] ubi: fastmap: Support for preseeded fastmap Richard Weinberger
2018-05-14 11:24 ` [PATCH 1/8] ubi: fastmap: Add support for flags Richard Weinberger
2018-05-14 11:24 ` [PATCH 2/8] ubi: fastmap: Add UBI_FM_SB_PRESEEDED_FLG flag Richard Weinberger
2018-05-14 11:24 ` [PATCH 3/8] ubi: fastmap: Implement PEB fixup Richard Weinberger
2018-05-14 11:24 ` [PATCH 4/8] ubi: fastmap: Handle bad block count for preseeded fastmap case Richard Weinberger
2018-05-14 11:24 ` [PATCH 5/8] ubi: fastmap: Fixup pool sizes for preseeded fastmaps Richard Weinberger
2018-05-14 11:24 ` Richard Weinberger [this message]
2018-05-14 11:24 ` [PATCH 7/8] ubi: fastmap: Relax size check Richard Weinberger
2018-05-14 11:24 ` [PATCH 8/8] ubi: Fix assert in ubi_wl_init Richard Weinberger

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=20180514112422.23988-7-richard@nod.at \
    --to=richard@nod.at \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mtd@lists.infradead.org \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).