Skip to contentSkip to conversation

net: stmmac: Premature loop termination check was ignored

PATCH In review Jochen Henneberg 4 participants 6 messages Feb 22, 2023 – Feb 23, 2023

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 net: stmmac: Premature loop termination check was ignored

The premature loop termination check makes sense only in case of the jump to read_again where the count may have been updated. But read_again did not include the check.

Fixes bba2556efad6 ("net: stmmac: Enable RX via AF_XDP zero-copy")
Fixes ec222003bd94 ("net: stmmac: Prepare to add Split Header support")

---
V2: Added fixes tags
V3: Fixed fixes tag format

 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +2 -2
index 1a5b8dab5e9b..de98c009866a 100644
@@ -5031,10 +5031,10 @@ static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue)
 			len = 0;
 		}
 
+read_again:
 		if (count >= limit)
 			break;
 
-read_again:
 		buf1_len = 0;
 		entry = next_entry;
 		buf = &rx_q->buf_pool[entry];
@@ -5221,10 +5221,10 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
 			len = 0;
 		}
 
+read_again:
 		if (count >= limit)
 			break;
 
-read_again:
 		buf1_len = 0;
 		buf2_len = 0;
 		entry = next_entry;
Signature · 1 line
2.39.2
Reviewed-by on v3 1/1
#1 lore raw
Quoted 6 lines · Jochen Henneberg

The premature loop termination check makes sense only in case of the jump to read_again where the count may have been updated. But read_again did not include the check.

Fixes bba2556efad6 ("net: stmmac: Enable RX via AF_XDP zero-copy")

This commit was included in v5.13

Quoted 1 line
Fixes ec222003bd94 ("net: stmmac: Prepare to add Split Header support")

While this one was included in v5.4

It seems to me that each of the above commits correspond to one of the two hunks below. I don't know if that means this patch should be split in two to assist backporting.

Quoted 1 line

That aside, this looks good to me.

Quoted 38 lines

---
V2: Added fixes tags
V3: Fixed fixes tag format

 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +2 -2
index 1a5b8dab5e9b..de98c009866a 100644
@@ -5031,10 +5031,10 @@ static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue)
 			len = 0;
 		}
 
+read_again:
 		if (count >= limit)
 			break;
 
-read_again:
 		buf1_len = 0;
 		entry = next_entry;
 		buf = &rx_q->buf_pool[entry];
@@ -5221,10 +5221,10 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
 			len = 0;
 		}
 
+read_again:
 		if (count >= limit)
 			break;
 
-read_again:
 		buf1_len = 0;
 		buf2_len = 0;
 		entry = next_entry;
Signature · 1 line
2.39.2
#2 lore raw
Quoted 10 lines · Simon Horman
Quoted 6 lines · Jochen Henneberg

The premature loop termination check makes sense only in case of the jump to read_again where the count may have been updated. But read_again did not include the check.

Fixes bba2556efad6 ("net: stmmac: Enable RX via AF_XDP zero-copy")

This commit was included in v5.13

Quoted 8 lines
Quoted 1 line
Fixes ec222003bd94 ("net: stmmac: Prepare to add Split Header support")

While this one was included in v5.4

It seems to me that each of the above commits correspond to one of the two hunks below. I don't know if that means this patch should be split in two to assist backporting.

I was thinking about this already but the change was so trivial that I hesitated to split it into two commits. I wanted I will surely change this.

Quoted 44 lines
Quoted 1 line

That aside, this looks good to me.

Quoted 38 lines

---
V2: Added fixes tags
V3: Fixed fixes tag format

 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +2 -2
index 1a5b8dab5e9b..de98c009866a 100644
@@ -5031,10 +5031,10 @@ static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue)
 			len = 0;
 		}
 
+read_again:
 		if (count >= limit)
 			break;
 
-read_again:
 		buf1_len = 0;
 		entry = next_entry;
 		buf = &rx_q->buf_pool[entry];
@@ -5221,10 +5221,10 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
 			len = 0;
 		}
 
+read_again:
 		if (count >= limit)
 			break;
 
-read_again:
 		buf1_len = 0;
 		buf2_len = 0;
 		entry = next_entry;
Signature · 1 line
2.39.2
Signature · 7 lines
Henneberg - Systemdesign
Jochen Henneberg
Loehnfeld 26
21423 Winsen (Luhe)
--
Fon: +49 172 160 14 69
Url: https://www.henneberg-systemdesign.com
#3 lore raw
Quoted 17 lines · Henneberg - Systemdesign
Quoted 14 lines · Simon Horman
Quoted 6 lines · Jochen Henneberg

The premature loop termination check makes sense only in case of the jump to read_again where the count may have been updated. But read_again did not include the check.

Fixes bba2556efad6 ("net: stmmac: Enable RX via AF_XDP zero-copy")

This commit was included in v5.13

Quoted 1 line
Fixes ec222003bd94 ("net: stmmac: Prepare to add Split Header support")

While this one was included in v5.4

Quoted 8 lines
Quoted 4 lines

It seems to me that each of the above commits correspond to one of the two hunks below. I don't know if that means this patch should be split in two to assist backporting.

I was thinking about this already but the change was so trivial that I hesitated to split it into two commits. I wanted I will surely change this.

The advantage of splitting is that it makes back porting easy. Both parts are needed for 6.1 and 5.15. 5.10 only needs the fix for ec222003bd94. It if does not easily apply to 5.10 it could get dropped. By splitting it, the backporting probably happens fully automated, no human involved.

Andrew

#4 lore raw
Quoted 27 lines · Andrew Lunn
Quoted 17 lines · Henneberg - Systemdesign
Quoted 14 lines · Simon Horman
Quoted 6 lines · Jochen Henneberg

The premature loop termination check makes sense only in case of the jump to read_again where the count may have been updated. But read_again did not include the check.

Fixes bba2556efad6 ("net: stmmac: Enable RX via AF_XDP zero-copy")

This commit was included in v5.13

Quoted 1 line
Fixes ec222003bd94 ("net: stmmac: Prepare to add Split Header support")

While this one was included in v5.4

Quoted 8 lines
Quoted 4 lines

It seems to me that each of the above commits correspond to one of the two hunks below. I don't know if that means this patch should be split in two to assist backporting.

I was thinking about this already but the change was so trivial that I hesitated to split it into two commits. I wanted I will surely change this.

Quoted 5 lines

The advantage of splitting is that it makes back porting easy. Both parts are needed for 6.1 and 5.15. 5.10 only needs the fix for ec222003bd94. It if does not easily apply to 5.10 it could get dropped. By splitting it, the backporting probably happens fully automated, no human involved.

Understood. Will do the split and send two new patches. I will not continue with version upcounting and not send a patch series but two completly independent patches.

Jochen

Quoted 2 lines

Andrew

Signature · 7 lines
Henneberg - Systemdesign
Jochen Henneberg
Loehnfeld 26
21423 Winsen (Luhe)
--
Fon: +49 172 160 14 69
Url: https://www.henneberg-systemdesign.com
#5 lore raw
Quoted 36 lines · Jochen Henneberg
Quoted 27 lines · Andrew Lunn
Quoted 17 lines · Henneberg - Systemdesign
Quoted 14 lines · Simon Horman
Quoted 6 lines · Jochen Henneberg

The premature loop termination check makes sense only in case of the jump to read_again where the count may have been updated. But read_again did not include the check.

Fixes bba2556efad6 ("net: stmmac: Enable RX via AF_XDP zero-copy")

This commit was included in v5.13

Quoted 1 line
Fixes ec222003bd94 ("net: stmmac: Prepare to add Split Header support")

While this one was included in v5.4

Quoted 8 lines
Quoted 4 lines

It seems to me that each of the above commits correspond to one of the two hunks below. I don't know if that means this patch should be split in two to assist backporting.

I was thinking about this already but the change was so trivial that I hesitated to split it into two commits. I wanted I will surely change this.

Quoted 5 lines

The advantage of splitting is that it makes back porting easy. Both parts are needed for 6.1 and 5.15. 5.10 only needs the fix for ec222003bd94. It if does not easily apply to 5.10 it could get dropped. By splitting it, the backporting probably happens fully automated, no human involved.

Quoted 3 lines

Understood. Will do the split and send two new patches. I will not continue with version upcounting and not send a patch series but two completly independent patches.

There may be fuzz if the patches are not applied in order. I'd suggest making a series.