erofs: address D-cache aliasing

Flush the D-cache before unlocking folios for compressed inodes, as
they are dirtied during decompression.

Avoid calling flush_dcache_folio() on every CPU write, since it's more
like playing whack-a-mole without real benefit.

It has no impact on x86 and arm64/risc-v: on x86, flush_dcache_folio()
is a no-op, and on arm64/risc-v, PG_dcache_clean (PG_arch_1) is clear
for new page cache folios.  However, certain ARM boards are affected,
as reported.

Fixes: 3883a79 ("staging: erofs: introduce VLE decompression support")
Closes: https://lore.kernel.org/r/c1e51e16-6cc6-49d0-a63e-4e9ff6c4dd53@pengutronix.de
Closes: https://lore.kernel.org/r/38d43fae-1182-4155-9c5b-ffc7382d9917@siemens.com
Tested-by: Jan Kiszka <jan.kiszka@siemens.com>
Tested-by: Stefan Kerkmann <s.kerkmann@pengutronix.de>
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Link: https://lore.kernel.org/r/20250709034614.2780117-2-hsiangkao@linux.alibaba.com

Author: hsiangkao

fs/erofs/data.c

@@ -214,9 +214,11 @@ int erofs_map_dev(struct super_block *sb, struct erofs_map_dev *map)
 
 /*
  * bit 30: I/O error occurred on this folio
+ * bit 29: CPU has dirty data in D-cache (needs aliasing handling);
  * bit 0 - 29: remaining parts to complete this folio
  */
-#define EROFS_ONLINEFOLIO_EIO			(1 << 30)
+#define EROFS_ONLINEFOLIO_EIO		30
+#define EROFS_ONLINEFOLIO_DIRTY		29
 
 void erofs_onlinefolio_init(struct folio *folio)
 {
@@ -233,19 +235,23 @@ void erofs_onlinefolio_split(struct folio *folio)
 	atomic_inc((atomic_t *)&folio->private);
 }
 
-void erofs_onlinefolio_end(struct folio *folio, int err)
+void erofs_onlinefolio_end(struct folio *folio, int err, bool dirty)
 {
 	int orig, v;
 
 	do {
 		orig = atomic_read((atomic_t *)&folio->private);
-		v = (orig - 1) | (err ? EROFS_ONLINEFOLIO_EIO : 0);
+		DBG_BUGON(orig <= 0);
+		v = dirty << EROFS_ONLINEFOLIO_DIRTY;
+		v |= (orig - 1) | (!!err << EROFS_ONLINEFOLIO_EIO);
 	} while (atomic_cmpxchg((atomic_t *)&folio->private, orig, v) != orig);
 
-	if (v & ~EROFS_ONLINEFOLIO_EIO)
+	if (v & (BIT(EROFS_ONLINEFOLIO_DIRTY) - 1))
 		return;
 	folio->private = 0;
-	folio_end_read(folio, !(v & EROFS_ONLINEFOLIO_EIO));
+	if (v & BIT(EROFS_ONLINEFOLIO_DIRTY))
+		flush_dcache_folio(folio);
+	folio_end_read(folio, !(v & BIT(EROFS_ONLINEFOLIO_EIO)));
 }
 
 static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,

fs/erofs/decompressor.c

@@ -301,13 +301,11 @@ static int z_erofs_transform_plain(struct z_erofs_decompress_req *rq,
 		cur = min(cur, rq->outputsize);
 		if (cur && rq->out[0]) {
 			kin = kmap_local_page(rq->in[nrpages_in - 1]);
-			if (rq->out[0] == rq->in[nrpages_in - 1]) {
+			if (rq->out[0] == rq->in[nrpages_in - 1])
 				memmove(kin + rq->pageofs_out, kin + pi, cur);
-				flush_dcache_page(rq->out[0]);
-			} else {
+			else
 				memcpy_to_page(rq->out[0], rq->pageofs_out,
 					       kin + pi, cur);
-			}
 			kunmap_local(kin);
 		}
 		rq->outputsize -= cur;
@@ -325,14 +323,12 @@ static int z_erofs_transform_plain(struct z_erofs_decompress_req *rq,
 			po = (rq->pageofs_out + cur + pi) & ~PAGE_MASK;
 			DBG_BUGON(no >= nrpages_out);
 			cnt = min(insz - pi, PAGE_SIZE - po);
-			if (rq->out[no] == rq->in[ni]) {
+			if (rq->out[no] == rq->in[ni])
 				memmove(kin + po,
 					kin + rq->pageofs_in + pi, cnt);
-				flush_dcache_page(rq->out[no]);
-			} else if (rq->out[no]) {
+			else if (rq->out[no])
 				memcpy_to_page(rq->out[no], po,
 					       kin + rq->pageofs_in + pi, cnt);
-			}
 			pi += cnt;
 		} while (pi < insz);
 		kunmap_local(kin);

fs/erofs/fileio.c

@@ -38,7 +38,7 @@ static void erofs_fileio_ki_complete(struct kiocb *iocb, long ret)
 	} else {
 		bio_for_each_folio_all(fi, &rq->bio) {
 			DBG_BUGON(folio_test_uptodate(fi.folio));
-			erofs_onlinefolio_end(fi.folio, ret);
+			erofs_onlinefolio_end(fi.folio, ret, false);
 		}
 	}
 	bio_uninit(&rq->bio);
@@ -154,7 +154,7 @@ static int erofs_fileio_scan_folio(struct erofs_fileio *io, struct folio *folio)
 		}
 		cur += len;
 	}
-	erofs_onlinefolio_end(folio, err);
+	erofs_onlinefolio_end(folio, err, false);
 	return err;
 }
 

fs/erofs/internal.h

@@ -390,7 +390,7 @@ int erofs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 int erofs_map_blocks(struct inode *inode, struct erofs_map_blocks *map);
 void erofs_onlinefolio_init(struct folio *folio);
 void erofs_onlinefolio_split(struct folio *folio);
-void erofs_onlinefolio_end(struct folio *folio, int err);
+void erofs_onlinefolio_end(struct folio *folio, int err, bool dirty);
 struct inode *erofs_iget(struct super_block *sb, erofs_nid_t nid);
 int erofs_getattr(struct mnt_idmap *idmap, const struct path *path,
 		  struct kstat *stat, u32 request_mask,

fs/erofs/zdata.c

@@ -1091,7 +1091,7 @@ static int z_erofs_scan_folio(struct z_erofs_frontend *f,
 			tight = (bs == PAGE_SIZE);
 		}
 	} while ((end = cur) > 0);
-	erofs_onlinefolio_end(folio, err);
+	erofs_onlinefolio_end(folio, err, false);
 	return err;
 }
 
@@ -1196,7 +1196,7 @@ static void z_erofs_fill_other_copies(struct z_erofs_backend *be, int err)
 			cur += len;
 		}
 		kunmap_local(dst);
-		erofs_onlinefolio_end(page_folio(bvi->bvec.page), err);
+		erofs_onlinefolio_end(page_folio(bvi->bvec.page), err, true);
 		list_del(p);
 		kfree(bvi);
 	}
@@ -1355,7 +1355,7 @@ static int z_erofs_decompress_pcluster(struct z_erofs_backend *be, int err)
 
 		DBG_BUGON(z_erofs_page_is_invalidated(page));
 		if (!z_erofs_is_shortlived_page(page)) {
-			erofs_onlinefolio_end(page_folio(page), err);
+			erofs_onlinefolio_end(page_folio(page), err, true);
 			continue;
 		}
 		if (pcl->algorithmformat != Z_EROFS_COMPRESSION_LZ4) {