Skip to contentSkip to conversation

remoteproc: Reset table_ptr in rproc_start() failure paths

PATCH In review Suman Anna 2 participants 3 messages Jul 27, 2018 – Jul 31, 2018 Remote processor (remoteproc)

lore
Share & tools
lore

Suman Anna

author
#0 lore raw

PATCH 1/1 remoteproc: Reset table_ptr in rproc_start() failure paths

Unwind the modified table_ptr and restore it to the local copy upon any subsequent failures in the rproc_start() function. This keeps the function to remain balanced on failures without the need to balance any modified variables elsewhere.

While at this, do some minor cleanup of the extra lines between the failure labels as well.

---
 drivers/remoteproc/remoteproc_core.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/remoteproc/remoteproc_core.c +4 -3
index eadff6ce2f7f..afef2d491c5b 100644
@@ -953,7 +953,7 @@ static int rproc_start(struct rproc *rproc, const struct firmware *fw)
 	if (ret) {
 		dev_err(dev, "failed to prepare subdevices for %s: %d\n",
 			rproc->name, ret);
-		return ret;
+		goto reset_table_ptr;
 	}
 
 	/* power up the remote processor */
@@ -979,10 +979,11 @@ static int rproc_start(struct rproc *rproc, const struct firmware *fw)
 
 stop_rproc:
 	rproc->ops->stop(rproc);
-
 unprepare_subdevices:
 	rproc_unprepare_subdevices(rproc);
-
+reset_table_ptr:
+	if (loaded_table)
+		rproc->table_ptr = rproc->cached_table;
 	return ret;
 }
 
Signature · 1 line
2.18.0
#1 lore raw
Quoted 5 lines · Suman Anna

Unwind the modified table_ptr and restore it to the local copy upon any subsequent failures in the rproc_start() function. This keeps the function to remain balanced on failures without the need to balance any modified variables elsewhere.

Good catch.

Quoted 22 lines

While at this, do some minor cleanup of the extra lines between the failure labels as well.

---
 drivers/remoteproc/remoteproc_core.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/remoteproc/remoteproc_core.c +1 -1
index eadff6ce2f7f..afef2d491c5b 100644
@@ -953,7 +953,7 @@ static int rproc_start(struct rproc *rproc, const struct firmware *fw)
 	if (ret) {
 		dev_err(dev, "failed to prepare subdevices for %s: %d\n",
 			rproc->name, ret);
-		return ret;
+		goto reset_table_ptr;
 	}
 
 	/* power up the remote processor */

drivers/remoteproc/remoteproc_core.c static int rproc_start(struct rproc *rproc, const struct firmware *fw) · line 985

Quoted 9 lines
drivers/remoteproc/remoteproc_core.c +2 -2
@@ -979,10 +979,11 @@ static int rproc_start(struct rproc *rproc, const struct firmware *fw)
 
 stop_rproc:
 	rproc->ops->stop(rproc);
-
 unprepare_subdevices:
 	rproc_unprepare_subdevices(rproc);
-
+reset_table_ptr:
+	if (loaded_table)

Regardless of us having a loaded_table it should have the same value as cached_table when we return - which might be NULL if we don't have a resource table.

So I applied this without the conditional, please object if I missed something.

Regards,
Bjorn

Quoted 7 lines
drivers/remoteproc/remoteproc_core.c +1 -0
+		rproc->table_ptr = rproc->cached_table;
 	return ret;
 }
 
Signature · 1 line
2.18.0
author
#2 lore raw
Quoted 46 lines · Bjorn Andersson
Quoted 5 lines · Suman Anna

Unwind the modified table_ptr and restore it to the local copy upon any subsequent failures in the rproc_start() function. This keeps the function to remain balanced on failures without the need to balance any modified variables elsewhere.

Good catch.

Quoted 22 lines

While at this, do some minor cleanup of the extra lines between the failure labels as well.

---
 drivers/remoteproc/remoteproc_core.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/remoteproc/remoteproc_core.c +1 -1
index eadff6ce2f7f..afef2d491c5b 100644
@@ -953,7 +953,7 @@ static int rproc_start(struct rproc *rproc, const struct firmware *fw)
 	if (ret) {
 		dev_err(dev, "failed to prepare subdevices for %s: %d\n",
 			rproc->name, ret);
-		return ret;
+		goto reset_table_ptr;
 	}
 
 	/* power up the remote processor */

drivers/remoteproc/remoteproc_core.c static int rproc_start(struct rproc *rproc, const struct firmware *fw) · line 985

Quoted 9 lines
drivers/remoteproc/remoteproc_core.c +2 -2
@@ -979,10 +979,11 @@ static int rproc_start(struct rproc *rproc, const struct firmware *fw)
 
 stop_rproc:
 	rproc->ops->stop(rproc);
-
 unprepare_subdevices:
 	rproc_unprepare_subdevices(rproc);
-
+reset_table_ptr:
+	if (loaded_table)

Regardless of us having a loaded_table it should have the same value as cached_table when we return - which might be NULL if we don't have a resource table.

Quoted 2 lines

So I applied this without the conditional, please object if I missed something.

Yeah, that's fine. I added the conditional only to keep it symmetric, it gets modified only if loaded_table was non-NULL in the first place, and is pointing to the cached_table otherwise.

regards
Suman

Quoted 11 lines

Regards,
Bjorn

Quoted 7 lines
drivers/remoteproc/remoteproc_core.c +1 -0
+		rproc->table_ptr = rproc->cached_table;
 	return ret;
 }
 
Signature · 1 line
2.18.0