Fix zfs_get_data access to files with wrong generation

If TX_WRITE is create on a file, and the file is later deleted and a new
directory is created on the same object id, it is possible that when
zil_commit happens, zfs_get_data will be called on the new directory.
This may result in panic as it tries to do range lock.

This patch fixes this issue by record the generation number during
zfs_log_write, so zfs_get_data can check if the object is valid.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chunwei Chen <david.chen@nutanix.com>
Closes #10593
Closes #11682

cmd/ztest/ztest.c

@@ -2163,8 +2163,8 @@ ztest_get_done(zgd_t *zgd, int error)
 }
 
 static int
-ztest_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb,
-    zio_t *zio)
+ztest_get_data(void *arg, uint64_t arg2, lr_write_t *lr, char *buf,
+    struct lwb *lwb, zio_t *zio)
 {
 	ztest_ds_t *zd = arg;
 	objset_t *os = zd->zd_os;

include/sys/zil.h

@@ -399,6 +399,7 @@ typedef struct itx {
 	void		*itx_callback_data; /* User data for the callback */
 	size_t		itx_size;	/* allocated itx structure size */
 	uint64_t	itx_oid;	/* object id */
+	uint64_t	itx_gen;	/* gen number for zfs_get_data */
 	lr_t		itx_lr;		/* common part of log record */
 	/* followed by type-specific part of lr_xx_t and its immediate data */
 } itx_t;
@@ -467,7 +468,7 @@ typedef int zil_parse_blk_func_t(zilog_t *zilog, const blkptr_t *bp, void *arg,
 typedef int zil_parse_lr_func_t(zilog_t *zilog, const lr_t *lr, void *arg,
     uint64_t txg);
 typedef int zil_replay_func_t(void *arg1, void *arg2, boolean_t byteswap);
-typedef int zil_get_data_t(void *arg, lr_write_t *lr, char *dbuf,
+typedef int zil_get_data_t(void *arg, uint64_t arg2, lr_write_t *lr, char *dbuf,
     struct lwb *lwb, zio_t *zio);
 
 extern int zil_parse(zilog_t *zilog, zil_parse_blk_func_t *parse_blk_func,

include/sys/zvol_impl.h

@@ -85,8 +85,8 @@ void zvol_log_truncate(zvol_state_t *zv, dmu_tx_t *tx, uint64_t off,
     uint64_t len, boolean_t sync);
 void zvol_log_write(zvol_state_t *zv, dmu_tx_t *tx, uint64_t offset,
     uint64_t size, int sync);
-int zvol_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb,
-    zio_t *zio);
+int zvol_get_data(void *arg, uint64_t arg2, lr_write_t *lr, char *buf,
+    struct lwb *lwb, zio_t *zio);
 int zvol_init_impl(void);
 void zvol_fini_impl(void);
 void zvol_wait_close(zvol_state_t *zv);

module/os/linux/zfs/zfs_vnops_os.c

@@ -521,7 +521,8 @@ static int zil_fault_io = 0;
  * Get data to generate a TX_WRITE intent log record.
  */
 int
-zfs_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio)
+zfs_get_data(void *arg, uint64_t gen, lr_write_t *lr, char *buf,
+    struct lwb *lwb, zio_t *zio)
 {
 	zfsvfs_t *zfsvfs = arg;
 	objset_t *os = zfsvfs->z_os;
@@ -532,6 +533,7 @@ zfs_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio)
 	dmu_buf_t *db;
 	zgd_t *zgd;
 	int error = 0;
+	uint64_t zp_gen;
 
 	ASSERT3P(lwb, !=, NULL);
 	ASSERT3P(zio, !=, NULL);
@@ -550,6 +552,16 @@ zfs_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio)
 		zfs_zrele_async(zp);
 		return (SET_ERROR(ENOENT));
 	}
+	/* check if generation number matches */
+	if (sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs), &zp_gen,
+	    sizeof (zp_gen)) != 0) {
+		zfs_zrele_async(zp);
+		return (SET_ERROR(EIO));
+	}
+	if (zp_gen != gen) {
+		zfs_zrele_async(zp);
+		return (SET_ERROR(ENOENT));
+	}
 
 	zgd = kmem_zalloc(sizeof (zgd_t), KM_SLEEP);
 	zgd->zgd_lwb = lwb;

module/zfs/zfs_log.c

@@ -540,6 +540,7 @@ zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
 	uint32_t blocksize = zp->z_blksz;
 	itx_wr_state_t write_state;
 	uintptr_t fsync_cnt;
+	uint64_t gen = 0;
 
 	if (zil_replaying(zilog, tx) || zp->z_unlinked ||
 	    zfs_xattr_owner_unlinked(zp)) {
@@ -562,6 +563,9 @@ zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
 		(void) tsd_set(zfs_fsyncer_key, (void *)(fsync_cnt - 1));
 	}
 
+	(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(ZTOZSB(zp)), &gen,
+	    sizeof (gen));
+
 	while (resid) {
 		itx_t *itx;
 		lr_write_t *lr;
@@ -609,6 +613,7 @@ zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
 		BP_ZERO(&lr->lr_blkptr);
 
 		itx->itx_private = ZTOZSB(zp);
+		itx->itx_gen = gen;
 
 		if (!(ioflag & (O_SYNC | O_DSYNC)) && (zp->z_sync_cnt == 0) &&
 		    (fsync_cnt == 0))

module/zfs/zil.c

@@ -1744,7 +1744,8 @@ cont:
 			 * completed after "lwb_write_zio" completed.
 			 */
 			error = zilog->zl_get_data(itx->itx_private,
-			    lrwb, dbuf, lwb, lwb->lwb_write_zio);
+			    itx->itx_gen, lrwb, dbuf, lwb,
+			    lwb->lwb_write_zio);
 
 			if (error == EIO) {
 				txg_wait_synced(zilog->zl_dmu_pool, txg);

module/zfs/zvol.c

@@ -673,7 +673,8 @@ zvol_get_done(zgd_t *zgd, int error)
  * Get data to generate a TX_WRITE intent log record.
  */
 int
-zvol_get_data(void *arg, lr_write_t *lr, char *buf, struct lwb *lwb, zio_t *zio)
+zvol_get_data(void *arg, uint64_t arg2, lr_write_t *lr, char *buf,
+    struct lwb *lwb, zio_t *zio)
 {
 	zvol_state_t *zv = arg;
 	uint64_t offset = lr->lr_offset;