fsverity: stop using PG_error to track error status

As a step towards freeing the PG_error flag for other uses, change ext4
and f2fs to stop using PG_error to track verity errors.  Instead, if a
verity error occurs, just mark the whole bio as failed.  The coarser
granularity isn't really a problem since it isn't any worse than what
the block layer provides, and errors from a multi-page readahead aren't
reported to applications unless a single-page read fails too.

f2fs supports compression, which makes the f2fs changes a bit more
complicated than desired, but the basic premise still works.

Note: there are still a few uses of PageError in f2fs, but they are on
the write path, so they are unrelated and this patch doesn't touch them.

Reviewed-by: Chao Yu <chao@kernel.org>
Acked-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20221129070401.156114-1-ebiggers@kernel.org

fs/ext4/readpage.c

@@ -75,14 +75,10 @@ static void __read_end_io(struct bio *bio)
 	bio_for_each_segment_all(bv, bio, iter_all) {
 		page = bv->bv_page;
 
-		/* PG_error was set if verity failed. */
+		if (bio->bi_status)
-		if (bio->bi_status || PageError(page)) {
 			ClearPageUptodate(page);
-			/* will re-read again later */
+		else
-			ClearPageError(page);
-		} else {
 			SetPageUptodate(page);
-		}
 		unlock_page(page);
 	}
 	if (bio->bi_private)

fs/f2fs/compress.c

@@ -1711,50 +1711,27 @@ static void f2fs_put_dic(struct decompress_io_ctx *dic, bool in_task)
 	}
 }
 
-/*
- * Update and unlock the cluster's pagecache pages, and release the reference to
- * the decompress_io_ctx that was being held for I/O completion.
- */
-static void __f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
-				bool in_task)
-{
-	int i;
-
-	for (i = 0; i < dic->cluster_size; i++) {
-		struct page *rpage = dic->rpages[i];
-
-		if (!rpage)
-			continue;
-
-		/* PG_error was set if verity failed. */
-		if (failed || PageError(rpage)) {
-			ClearPageUptodate(rpage);
-			/* will re-read again later */
-			ClearPageError(rpage);
-		} else {
-			SetPageUptodate(rpage);
-		}
-		unlock_page(rpage);
-	}
-
-	f2fs_put_dic(dic, in_task);
-}
-
 static void f2fs_verify_cluster(struct work_struct *work)
 {
 	struct decompress_io_ctx *dic =
 		container_of(work, struct decompress_io_ctx, verity_work);
 	int i;
 
-	/* Verify the cluster's decompressed pages with fs-verity. */
+	/* Verify, update, and unlock the decompressed pages. */
 	for (i = 0; i < dic->cluster_size; i++) {
 		struct page *rpage = dic->rpages[i];
 
-		if (rpage && !fsverity_verify_page(rpage))
+		if (!rpage)
-			SetPageError(rpage);
+			continue;
+
+		if (fsverity_verify_page(rpage))
+			SetPageUptodate(rpage);
+		else
+			ClearPageUptodate(rpage);
+		unlock_page(rpage);
 	}
 
-	__f2fs_decompress_end_io(dic, false, true);
+	f2fs_put_dic(dic, true);
 }
 
 /*
@@ -1764,6 +1741,8 @@ static void f2fs_verify_cluster(struct work_struct *work)
 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
 				bool in_task)
 {
+	int i;
+
 	if (!failed && dic->need_verity) {
 		/*
 		 * Note that to avoid deadlocks, the verity work can't be done
@@ -1773,9 +1752,28 @@ void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
 		 */
 		INIT_WORK(&dic->verity_work, f2fs_verify_cluster);
 		fsverity_enqueue_verify_work(&dic->verity_work);
-	} else {
+		return;
-		__f2fs_decompress_end_io(dic, failed, in_task);
 	}
+
+	/* Update and unlock the cluster's pagecache pages. */
+	for (i = 0; i < dic->cluster_size; i++) {
+		struct page *rpage = dic->rpages[i];
+
+		if (!rpage)
+			continue;
+
+		if (failed)
+			ClearPageUptodate(rpage);
+		else
+			SetPageUptodate(rpage);
+		unlock_page(rpage);
+	}
+
+	/*
+	 * Release the reference to the decompress_io_ctx that was being held
+	 * for I/O completion.
+	 */
+	f2fs_put_dic(dic, in_task);
 }
 
 /*

fs/f2fs/data.c

@@ -116,43 +116,56 @@ struct bio_post_read_ctx {
 	struct f2fs_sb_info *sbi;
 	struct work_struct work;
 	unsigned int enabled_steps;
+	/*
+	 * decompression_attempted keeps track of whether
+	 * f2fs_end_read_compressed_page() has been called on the pages in the
+	 * bio that belong to a compressed cluster yet.
+	 */
+	bool decompression_attempted;
 	block_t fs_blkaddr;
 };
 
+/*
+ * Update and unlock a bio's pages, and free the bio.
+ *
+ * This marks pages up-to-date only if there was no error in the bio (I/O error,
+ * decryption error, or verity error), as indicated by bio->bi_status.
+ *
+ * "Compressed pages" (pagecache pages backed by a compressed cluster on-disk)
+ * aren't marked up-to-date here, as decompression is done on a per-compression-
+ * cluster basis rather than a per-bio basis.  Instead, we only must do two
+ * things for each compressed page here: call f2fs_end_read_compressed_page()
+ * with failed=true if an error occurred before it would have normally gotten
+ * called (i.e., I/O error or decryption error, but *not* verity error), and
+ * release the bio's reference to the decompress_io_ctx of the page's cluster.
+ */
 static void f2fs_finish_read_bio(struct bio *bio, bool in_task)
 {
 	struct bio_vec *bv;
 	struct bvec_iter_all iter_all;
+	struct bio_post_read_ctx *ctx = bio->bi_private;
 
-	/*
-	 * Update and unlock the bio's pagecache pages, and put the
-	 * decompression context for any compressed pages.
-	 */
 	bio_for_each_segment_all(bv, bio, iter_all) {
 		struct page *page = bv->bv_page;
 
 		if (f2fs_is_compressed_page(page)) {
-			if (bio->bi_status)
+			if (ctx && !ctx->decompression_attempted)
 				f2fs_end_read_compressed_page(page, true, 0,
 							in_task);
 			f2fs_put_page_dic(page, in_task);
 			continue;
 		}
 
-		/* PG_error was set if verity failed. */
+		if (bio->bi_status)
-		if (bio->bi_status || PageError(page)) {
 			ClearPageUptodate(page);
-			/* will re-read again later */
+		else
-			ClearPageError(page);
-		} else {
 			SetPageUptodate(page);
-		}
 		dec_page_count(F2FS_P_SB(page), __read_io_type(page));
 		unlock_page(page);
 	}
 
-	if (bio->bi_private)
+	if (ctx)
-		mempool_free(bio->bi_private, bio_post_read_ctx_pool);
+		mempool_free(ctx, bio_post_read_ctx_pool);
 	bio_put(bio);
 }
 
@@ -185,8 +198,10 @@ static void f2fs_verify_bio(struct work_struct *work)
 			struct page *page = bv->bv_page;
 
 			if (!f2fs_is_compressed_page(page) &&
-			    !fsverity_verify_page(page))
+			    !fsverity_verify_page(page)) {
-				SetPageError(page);
+				bio->bi_status = BLK_STS_IOERR;
+				break;
+			}
 		}
 	} else {
 		fsverity_verify_bio(bio);
@@ -245,6 +260,8 @@ static void f2fs_handle_step_decompress(struct bio_post_read_ctx *ctx,
 		blkaddr++;
 	}
 
+	ctx->decompression_attempted = true;
+
 	/*
 	 * Optimization: if all the bio's pages are compressed, then scheduling
 	 * the per-bio verity work is unnecessary, as verity will be fully
@@ -1062,6 +1079,7 @@ static struct bio *f2fs_grab_read_bio(struct inode *inode, block_t blkaddr,
 		ctx->sbi = sbi;
 		ctx->enabled_steps = post_read_steps;
 		ctx->fs_blkaddr = blkaddr;
+		ctx->decompression_attempted = false;
 		bio->bi_private = ctx;
 	}
 	iostat_alloc_and_bind_ctx(sbi, bio, ctx);
@@ -1089,7 +1107,6 @@ static int f2fs_submit_page_read(struct inode *inode, struct page *page,
 		bio_put(bio);
 		return -EFAULT;
 	}
-	ClearPageError(page);
 	inc_page_count(sbi, F2FS_RD_DATA);
 	f2fs_update_iostat(sbi, NULL, FS_DATA_READ_IO, F2FS_BLKSIZE);
 	__submit_bio(sbi, bio, DATA);
@@ -2141,7 +2158,6 @@ static int f2fs_read_single_page(struct inode *inode, struct page *page,
 	inc_page_count(F2FS_I_SB(inode), F2FS_RD_DATA);
 	f2fs_update_iostat(F2FS_I_SB(inode), NULL, FS_DATA_READ_IO,
 							F2FS_BLKSIZE);
-	ClearPageError(page);
 	*last_block_in_bio = block_nr;
 	goto out;
 out:
@@ -2289,7 +2305,6 @@ int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
 
 		inc_page_count(sbi, F2FS_RD_DATA);
 		f2fs_update_iostat(sbi, inode, FS_DATA_READ_IO, F2FS_BLKSIZE);
-		ClearPageError(page);
 		*last_block_in_bio = blkaddr;
 	}
 
@@ -2306,7 +2321,6 @@ int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
 	for (i = 0; i < cc->cluster_size; i++) {
 		if (cc->rpages[i]) {
 			ClearPageUptodate(cc->rpages[i]);
-			ClearPageError(cc->rpages[i]);
 			unlock_page(cc->rpages[i]);
 		}
 	}
@@ -2403,7 +2417,6 @@ static int f2fs_mpage_readpages(struct inode *inode,
 #ifdef CONFIG_F2FS_FS_COMPRESSION
 set_error_page:
 #endif
-			SetPageError(page);
 			zero_user_segment(page, 0, PAGE_SIZE);
 			unlock_page(page);
 		}

fs/verity/verify.c

@@ -200,9 +200,8 @@ EXPORT_SYMBOL_GPL(fsverity_verify_page);
  * @bio: the bio to verify
  *
  * Verify a set of pages that have just been read from a verity file.  The pages
- * must be pagecache pages that are still locked and not yet uptodate.  Pages
+ * must be pagecache pages that are still locked and not yet uptodate.  If a
- * that fail verification are set to the Error state.  Verification is skipped
+ * page fails verification, then bio->bi_status is set to an error status.
- * for pages already in the Error state, e.g. due to fscrypt decryption failure.
  *
  * This is a helper function for use by the ->readahead() method of filesystems
  * that issue bios to read data directly into the page cache.  Filesystems that
@@ -244,9 +243,10 @@ void fsverity_verify_bio(struct bio *bio)
 		unsigned long level0_ra_pages =
 			min(max_ra_pages, params->level0_blocks - level0_index);
 
-		if (!PageError(page) &&
+		if (!verify_page(inode, vi, req, page, level0_ra_pages)) {
-		    !verify_page(inode, vi, req, page, level0_ra_pages))
+			bio->bi_status = BLK_STS_IOERR;
-			SetPageError(page);
+			break;
+		}
 	}
 
 	fsverity_free_hash_request(params->hash_alg, req);