public inbox for linux-erofs@ozlabs.org
 help / color / mirror / Atom feed
From: Nithurshen <nithurshen.dev@gmail.com>
To: linux-erofs@lists.ozlabs.org
Cc: xiang@kernel.org, hsiangkao@linux.alibaba.com,
	Nithurshen <nithurshen.dev@gmail.com>
Subject: [PATCH] mkfs: support block map for blob devices
Date: Sat,  7 Mar 2026 11:58:10 +0530	[thread overview]
Message-ID: <20260307062810.19862-1-nithurshen.dev@gmail.com> (raw)

Currently, using --blobdev to specify an extra device is restricted from working with the block map chunk format. This was previously noted as a task that could be implemented by mapping the device blocks using a global address.

This patch implements this support by allowing the block map to reference chunks residing on extra devices. This is achieved by:
	1) Removing the -EINVAL check in mkfs/main.c that blocked this combination of flags.
	2) Calculating the global startblk address for the block map by summing the blocks of the primary device and any preceding extra devices.
	3) Ensuring that EROFS_CHUNK_FORMAT_INDEXES is only set if the user has not forced the block map format, and adjusting the index unit size to EROFS_BLOCK_MAP_ENTRY_SIZE accordingly.
	4) Updating erofs_inode_fixup_chunkformat to correctly identify when 48-bit addressing is required for a block-mapped file on an extra device.

Signed-off-by: Nithurshen <nithurshen.dev@gmail.com>
---
 lib/blobchunk.c | 34 ++++++++++++++++++++++++++++------
 mkfs/main.c     |  7 -------
 2 files changed, 28 insertions(+), 13 deletions(-)

diff --git a/lib/blobchunk.c b/lib/blobchunk.c
index 96c161b..2ef7462 100644
--- a/lib/blobchunk.c
+++ b/lib/blobchunk.c
@@ -159,7 +159,17 @@ void erofs_inode_fixup_chunkformat(struct erofs_inode *inode)
 		if (chunk->blkaddr == EROFS_NULL_ADDR)
 			continue;
 		if (chunk->device_id) {
-			if (chunk->blkaddr > UINT32_MAX) {
+			if (!(inode->u.chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
+				erofs_blk_t mapped_blkaddr = inode->sbi->primarydevice_blocks;
+				unsigned int i;
+
+				for (i = 0; i < chunk->device_id - 1; i++)
+					mapped_blkaddr += inode->sbi->devs[i].blocks;
+				if (mapped_blkaddr + chunk->blkaddr > UINT32_MAX) {
+					_48bit = true;
+					break;
+				}
+			} else if (chunk->blkaddr > UINT32_MAX) {
 				_48bit = true;
 				break;
 			}
@@ -201,8 +211,16 @@ int erofs_write_chunk_indexes(struct erofs_inode *inode, struct erofs_vfile *vf,
 		if (chunk->blkaddr == EROFS_NULL_ADDR) {
 			startblk = EROFS_NULL_ADDR;
 		} else if (chunk->device_id) {
-			DBG_BUGON(!(inode->u.chunkformat & EROFS_CHUNK_FORMAT_INDEXES));
-			startblk = chunk->blkaddr;
+			if (inode->u.chunkformat & EROFS_CHUNK_FORMAT_INDEXES) {
+				startblk = chunk->blkaddr;
+			} else {
+				unsigned int i;
+
+				startblk = sbi->primarydevice_blocks;
+				for (i = 0; i < chunk->device_id - 1; i++)
+					startblk += sbi->devs[i].blocks;
+				startblk += chunk->blkaddr;
+			}
 			extent_start = EROFS_NULL_ADDR;
 		} else {
 			startblk = remapped_base + chunk->blkaddr;
@@ -324,7 +342,7 @@ int erofs_blob_write_chunked_file(struct erofs_inode *inode, int fd,
 	chunksize = 1ULL << chunkbits;
 	count = DIV_ROUND_UP(inode->i_size, chunksize);
 
-	if (sbi->extra_devices)
+	if (sbi->extra_devices && cfg.c_force_chunkformat != FORCE_INODE_BLOCK_MAP)
 		inode->u.chunkformat |= EROFS_CHUNK_FORMAT_INDEXES;
 	if (inode->u.chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
 		unit = sizeof(struct erofs_inode_chunk_index);
@@ -494,8 +512,12 @@ int tarerofs_write_chunkes(struct erofs_inode *inode, erofs_off_t data_offset)
 	inode->u.chunkformat |= chunkbits - sbi->blkszbits;
 	if (sbi->extra_devices) {
 		device_id = 1;
-		inode->u.chunkformat |= EROFS_CHUNK_FORMAT_INDEXES;
-		unit = sizeof(struct erofs_inode_chunk_index);
+		if (cfg.c_force_chunkformat != FORCE_INODE_BLOCK_MAP)
+			inode->u.chunkformat |= EROFS_CHUNK_FORMAT_INDEXES;
+		if (inode->u.chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
+			unit = sizeof(struct erofs_inode_chunk_index);
+		else
+			unit = EROFS_BLOCK_MAP_ENTRY_SIZE;
 		DBG_BUGON(erofs_blkoff(sbi, data_offset));
 		blkaddr = erofs_blknr(sbi, data_offset);
 	} else {
diff --git a/mkfs/main.c b/mkfs/main.c
index 58c18f9..07ef086 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -1565,13 +1565,6 @@ static int mkfs_parse_options_cfg(struct erofs_importer_params *params,
 		return -EINVAL;
 	}
 
-	/* TODO: can be implemented with (deviceslot) mapped_blkaddr */
-	if (cfg.c_blobdev_path &&
-	    cfg.c_force_chunkformat == FORCE_INODE_BLOCK_MAP) {
-		erofs_err("--blobdev cannot work with block map currently");
-		return -EINVAL;
-	}
-
 	if (optind >= argc) {
 		erofs_err("missing argument: FILE");
 		return -EINVAL;
-- 
2.51.0



             reply	other threads:[~2026-03-07  6:28 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-07  6:28 Nithurshen [this message]
2026-03-09 11:38 ` [PATCH] mkfs: support block map for blob devices zhaoyifan (H)
2026-03-09 13:01   ` Nithurshen Karthikeyan
2026-03-12  9:56   ` Nithurshen Karthikeyan
2026-03-12 11:40     ` Gao Xiang
2026-03-12 13:21       ` Nithurshen Karthikeyan
2026-03-19  3:19       ` Nithurshen Karthikeyan
2026-03-09 12:24 ` [PATCH v2] " Nithurshen

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=20260307062810.19862-1-nithurshen.dev@gmail.com \
    --to=nithurshen.dev@gmail.com \
    --cc=hsiangkao@linux.alibaba.com \
    --cc=linux-erofs@lists.ozlabs.org \
    --cc=xiang@kernel.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