On 2021/6/11 下午3:49, Johannes Thumshirn wrote:

On 11/06/2021 03:32, Qu Wenruo wrote:

quoted

+		/* Allocate new bio if not allocated or already submitted */
+		if (!bio) {
+			bio = alloc_compressed_bio(cb, cur_disk_bytenr,
+				bio_op | write_flags,
+				end_compressed_bio_write,
+				&next_stripe_start);
+			if (IS_ERR(bio)) {
+				ret = errno_to_blk_status(PTR_ERR(bio));
+				bio = NULL;
+				goto finish_cb;
+			}
+		}
+		/*
+		 * We should never reach next_stripe_start, as if we reach the
+		 * boundary we will submit the bio immediately.
+		 */
+		ASSERT(cur_disk_bytenr != next_stripe_start);
+
+		/*
+		 * We have various limit on the real read size:
+		 * - stripe boundary
+		 * - page boundary
+		 * - compressed length boundary
+		 */
+		real_size = min_t(u64, U32_MAX,
+				  next_stripe_start - cur_disk_bytenr);
+		real_size = min_t(u64, real_size,
+				  PAGE_SIZE - offset_in_page(offset));
+		real_size = min_t(u64, real_size,
+				  compressed_len - offset);
+		ASSERT(IS_ALIGNED(real_size, fs_info->sectorsize));

-	/* create and submit bios for the compressed pages */
-	bytes_left = compressed_len;
-	for (pg_index = 0; pg_index < cb->nr_pages; pg_index++) {
-		int submit = 0;
-		int len;
+		added = bio_add_page(bio, page, real_size,
+				     offset_in_page(offset));
+		/*
+		 * Maximum compressed extent size is 128K, we should never
+		 * reach bio size limit.
+		 */
+		ASSERT(added == real_size);

-		page = compressed_pages[pg_index];
-		page->mapping = inode->vfs_inode.i_mapping;
-		if (bio->bi_iter.bi_size)
-			submit = btrfs_bio_fits_in_stripe(page, PAGE_SIZE, bio,
-							  0);
+		cur_disk_bytenr += added;

-		if (pg_index == 0 && use_append)
-			len = bio_add_zone_append_page(bio, page, PAGE_SIZE, 0);
-		else
-			len = bio_add_page(bio, page, PAGE_SIZE, 0);

I think you still need to distinguish between normal write and zone append here,
as you adding pages to an already created bio.

Oh, my bad, forgot to handle the zoned append differently.

Adding one page to an empty bio
will always succeed but when adding more than one page to a zone append bio, you
have to take the device's maximum zone append limit into account, as zone append
bios can't be split. This is also the reason why we do the device
lookup/bio_set_dev() for the zone append bios, so bio_add_zone_append_page() can
look at the device's limitations when adding the pages.

Did you mean that for the bio_add_zone_append_page(), it may return less
bytes than we expected?
Even if our compressed write is ensured to be smaller than 128K?

Thanks,
Qu