Skip to contentSkip to conversation

ipc/sem: Remove redundant assignments

PATCH Applied Michal Orzel 3 participants 9 messages Apr 9, 2022 – Apr 12, 2022

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ipc/sem: Remove redundant assignments

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

Reported by clang-tidy [deadcode.DeadStores]

---
 ipc/sem.c | 2 --
 1 file changed, 2 deletions(-)
ipc/sem.c +0 -2
index 0dbdb98fdf2d..38ef91a63edd 100644
@@ -766,7 +766,6 @@ static int perform_atomic_semop(struct sem_array *sma, struct sem_queue *q)
 	for (sop = sops; sop < sops + nsops; sop++) {
 		curr = &sma->sems[sop->sem_num];
 		sem_op = sop->sem_op;
-		result = curr->semval;
 
 		if (sop->sem_flg & SEM_UNDO) {
 			int undo = un->semadj[sop->sem_num] - sem_op;
@@ -1430,7 +1429,6 @@ static int semctl_main(struct ipc_namespace *ns, int semid, int semnum,
 	if (err)
 		goto out_rcu_wakeup;
 
-	err = -EACCES;
 	switch (cmd) {
 	case GETALL:
 	{
Signature · 1 line
2.25.1
#1 lore raw

PATCH 1/1 block: Remove redundant assignments

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

Reported by clang-tidy [deadcode.DeadStores]

---
 block/badblocks.c        |  2 --
 block/blk-map.c          |  5 ++---
 block/partitions/acorn.c |  4 ++--
 block/partitions/atari.c |  1 -
 block/partitions/ldm.c   | 15 +++------------
 5 files changed, 7 insertions(+), 20 deletions(-)
block/badblocks.c +0 -2
index d39056630d9c..3afb550c0f7b 100644
@@ -65,7 +65,6 @@ int badblocks_check(struct badblocks *bb, sector_t s, int sectors,
 		s >>= bb->shift;
 		target += (1<<bb->shift) - 1;
 		target >>= bb->shift;
-		sectors = target - s;
 	}
 	/* 'target' is now the first block after the bad range */
 
@@ -345,7 +344,6 @@ int badblocks_clear(struct badblocks *bb, sector_t s, int sectors)
 		s += (1<<bb->shift) - 1;
 		s >>= bb->shift;
 		target >>= bb->shift;
-		sectors = target - s;
 	}
 
 	write_seqlock_irq(&bb->lock);
block/blk-map.c +2 -3
index c7f71d83eff1..fa72e63e18c2 100644
@@ -260,10 +260,9 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,
 
 		npages = DIV_ROUND_UP(offs + bytes, PAGE_SIZE);
 
-		if (unlikely(offs & queue_dma_alignment(rq->q))) {
-			ret = -EINVAL;
+		if (unlikely(offs & queue_dma_alignment(rq->q)))
 			j = 0;
-		} else {
+		else {
 			for (j = 0; j < npages; j++) {
 				struct page *page = pages[j];
 				unsigned int n = PAGE_SIZE - offs;
block/partitions/acorn.c +2 -2
index 2c381c694c57..d2fc122d7426 100644
@@ -282,13 +282,13 @@ int adfspart_check_ADFS(struct parsed_partitions *state)
 #ifdef CONFIG_ACORN_PARTITION_RISCIX
 		case PARTITION_RISCIX_SCSI:
 		case PARTITION_RISCIX_MFM:
-			slot = riscix_partition(state, start_sect, slot,
+			riscix_partition(state, start_sect, slot,
 						nr_sects);
 			break;
 #endif
 
 		case PARTITION_LINUX:
-			slot = linux_partition(state, start_sect, slot,
+			linux_partition(state, start_sect, slot,
 					       nr_sects);
 			break;
 		}
block/partitions/atari.c +0 -1
index da5994175416..9655c728262a 100644
@@ -140,7 +140,6 @@ int atari_partition(struct parsed_partitions *state)
 				/* accept only GEM,BGM,RAW,LNX,SWP partitions */
 				if (!((pi->flg & 1) && OK_id(pi->id)))
 					continue;
-				part_fmt = 2;
 				put_partition (state, slot,
 						be32_to_cpu(pi->st),
 						be32_to_cpu(pi->siz));
block/partitions/ldm.c +3 -12
index 27f6c7d9c776..38e58960ae03 100644
@@ -736,7 +736,6 @@ static bool ldm_parse_cmp3 (const u8 *buffer, int buflen, struct vblk *vb)
 		len = r_cols;
 	} else {
 		r_stripe = 0;
-		r_cols   = 0;
 		len = r_parent;
 	}
 	if (len < 0)
@@ -783,11 +782,8 @@ static int ldm_parse_dgr3 (const u8 *buffer, int buflen, struct vblk *vb)
 		r_id1 = ldm_relative (buffer, buflen, 0x24, r_diskid);
 		r_id2 = ldm_relative (buffer, buflen, 0x24, r_id1);
 		len = r_id2;
-	} else {
-		r_id1 = 0;
-		r_id2 = 0;
+	} else
 		len = r_diskid;
-	}
 	if (len < 0)
 		return false;
 
@@ -826,11 +822,8 @@ static bool ldm_parse_dgr4 (const u8 *buffer, int buflen, struct vblk *vb)
 		r_id1 = ldm_relative (buffer, buflen, 0x44, r_name);
 		r_id2 = ldm_relative (buffer, buflen, 0x44, r_id1);
 		len = r_id2;
-	} else {
-		r_id1 = 0;
-		r_id2 = 0;
+	} else
 		len = r_name;
-	}
 	if (len < 0)
 		return false;
 
@@ -963,10 +956,8 @@ static bool ldm_parse_prt3(const u8 *buffer, int buflen, struct vblk *vb)
 			return false;
 		}
 		len = r_index;
-	} else {
-		r_index = 0;
+	} else
 		len = r_diskid;
-	}
 	if (len < 0) {
 		ldm_error("len %d < 0", len);
 		return false;
Signature · 1 line
2.25.1
#2 lore raw

Re: [PATCH] block: Remove redundant assignments

Quoted 2 lines · Michal Orzel

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

This log is the same as your last patch.

Instead of a general statement on deadstores, a more specific

analysis of the setting being removed would be helpful.

This will mean splitting the patch to match the analysis.

Tom

Quoted 131 lines

Reported by clang-tidy [deadcode.DeadStores]

---
  block/badblocks.c        |  2 --
  block/blk-map.c          |  5 ++---
  block/partitions/acorn.c |  4 ++--
  block/partitions/atari.c |  1 -
  block/partitions/ldm.c   | 15 +++------------
  5 files changed, 7 insertions(+), 20 deletions(-)
block/badblocks.c +0 -2
index d39056630d9c..3afb550c0f7b 100644
@@ -65,7 +65,6 @@ int badblocks_check(struct badblocks *bb, sector_t s, int sectors,
  		s >>= bb->shift;
  		target += (1<<bb->shift) - 1;
  		target >>= bb->shift;
-		sectors = target - s;
  	}
  	/* 'target' is now the first block after the bad range */
 
@@ -345,7 +344,6 @@ int badblocks_clear(struct badblocks *bb, sector_t s, int sectors)
  		s += (1<<bb->shift) - 1;
  		s >>= bb->shift;
  		target >>= bb->shift;
-		sectors = target - s;
  	}
 
  	write_seqlock_irq(&bb->lock);
block/blk-map.c +2 -3
index c7f71d83eff1..fa72e63e18c2 100644
@@ -260,10 +260,9 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,
 
  		npages = DIV_ROUND_UP(offs + bytes, PAGE_SIZE);
 
-		if (unlikely(offs & queue_dma_alignment(rq->q))) {
-			ret = -EINVAL;
+		if (unlikely(offs & queue_dma_alignment(rq->q)))
  			j = 0;
-		} else {
+		else {
  			for (j = 0; j < npages; j++) {
  				struct page *page = pages[j];
  				unsigned int n = PAGE_SIZE - offs;
block/partitions/acorn.c +2 -2
index 2c381c694c57..d2fc122d7426 100644
@@ -282,13 +282,13 @@ int adfspart_check_ADFS(struct parsed_partitions *state)
  #ifdef CONFIG_ACORN_PARTITION_RISCIX
  		case PARTITION_RISCIX_SCSI:
  		case PARTITION_RISCIX_MFM:
-			slot = riscix_partition(state, start_sect, slot,
+			riscix_partition(state, start_sect, slot,
  						nr_sects);
  			break;
  #endif
 
  		case PARTITION_LINUX:
-			slot = linux_partition(state, start_sect, slot,
+			linux_partition(state, start_sect, slot,
  					       nr_sects);
  			break;
  		}
block/partitions/atari.c +0 -1
index da5994175416..9655c728262a 100644
@@ -140,7 +140,6 @@ int atari_partition(struct parsed_partitions *state)
  				/* accept only GEM,BGM,RAW,LNX,SWP partitions */
  				if (!((pi->flg & 1) && OK_id(pi->id)))
  					continue;
-				part_fmt = 2;
  				put_partition (state, slot,
  						be32_to_cpu(pi->st),
  						be32_to_cpu(pi->siz));
block/partitions/ldm.c +3 -12
index 27f6c7d9c776..38e58960ae03 100644
@@ -736,7 +736,6 @@ static bool ldm_parse_cmp3 (const u8 *buffer, int buflen, struct vblk *vb)
  		len = r_cols;
  	} else {
  		r_stripe = 0;
-		r_cols   = 0;
  		len = r_parent;
  	}
  	if (len < 0)
@@ -783,11 +782,8 @@ static int ldm_parse_dgr3 (const u8 *buffer, int buflen, struct vblk *vb)
  		r_id1 = ldm_relative (buffer, buflen, 0x24, r_diskid);
  		r_id2 = ldm_relative (buffer, buflen, 0x24, r_id1);
  		len = r_id2;
-	} else {
-		r_id1 = 0;
-		r_id2 = 0;
+	} else
  		len = r_diskid;
-	}
  	if (len < 0)
  		return false;
 
@@ -826,11 +822,8 @@ static bool ldm_parse_dgr4 (const u8 *buffer, int buflen, struct vblk *vb)
  		r_id1 = ldm_relative (buffer, buflen, 0x44, r_name);
  		r_id2 = ldm_relative (buffer, buflen, 0x44, r_id1);
  		len = r_id2;
-	} else {
-		r_id1 = 0;
-		r_id2 = 0;
+	} else
  		len = r_name;
-	}
  	if (len < 0)
  		return false;
 
@@ -963,10 +956,8 @@ static bool ldm_parse_prt3(const u8 *buffer, int buflen, struct vblk *vb)
  			return false;
  		}
  		len = r_index;
-	} else {
-		r_index = 0;
+	} else
  		len = r_diskid;
-	}
  	if (len < 0) {
  		ldm_error("len %d < 0", len);
  		return false;
#3 lore raw

Re: [PATCH] block: Remove redundant assignments

Hi Tom,

Quoted 7 lines · Michal Orzel
Quoted 2 lines

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

This log is the same as your last patch.

Quoted 8 lines

Instead of a general statement on deadstores, a more specific

analysis of the setting being removed would be helpful.

This will mean splitting the patch to match the analysis.

Tom

I think that this explanation in a commit is enough for such a trivial patch. It informs that we are fixing clang-tidy findings related to deadstores
in a block subsystem. What analysis would you want to see?
Something like "Remove deadstore assignment ret = -EINVAL in a function bio_map_user_iov" ? I think that it will create too much of not needed overhead.

Cheers,
Michal

#4 lore raw

Re: [PATCH] block: Remove redundant assignments

Quoted 13 lines · Michal Orzel

Hi Tom,

Quoted 11 lines · Michal Orzel
Quoted 2 lines

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

This log is the same as your last patch.

Instead of a general statement on deadstores, a more specific

analysis of the setting being removed would be helpful.

This will mean splitting the patch to match the analysis.

Quoted 7 lines
Quoted 2 lines

Tom

I think that this explanation in a commit is enough for such a trivial patch. It informs that we are fixing clang-tidy findings related to deadstores
in a block subsystem. What analysis would you want to see?
Something like "Remove deadstore assignment ret = -EINVAL in a function bio_map_user_iov" ? I think that it will create too much of not needed overhead.

The compiler should remove these already.  All of the static analyzers produce false positives, if you do not provide analysis the reviewer has to assume you have not done any and has to do the false positive analysis themselves. This burdens the reviewer for something of small value that can be dismissed as churn.

Tom

Quoted 4 lines

Cheers,
Michal

#5 lore raw

Re: [PATCH] block: Remove redundant assignments

Hi Tom,

Quoted 23 lines · Tom Rix
Quoted 13 lines · Michal Orzel

Hi Tom,

Quoted 11 lines · Michal Orzel
Quoted 2 lines

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

This log is the same as your last patch.

Instead of a general statement on deadstores, a more specific

analysis of the setting being removed would be helpful.

This will mean splitting the patch to match the analysis.

Quoted 7 lines
Quoted 2 lines

Tom

I think that this explanation in a commit is enough for such a trivial patch. It informs that we are fixing clang-tidy findings related to deadstores
in a block subsystem. What analysis would you want to see?
Something like "Remove deadstore assignment ret = -EINVAL  in a function bio_map_user_iov" ? I think that it will create too much of not needed overhead.

Quoted 4 lines

The compiler should remove these already.  All of the static analyzers produce false positives, if you do not provide analysis the reviewer has to assume you have not done any and has to do the false positive analysis themselves. This burdens the reviewer for something of small value that can be dismissed as churn.

Tom

Ok, I will split the patch into series in v2 and I will add some analysis into commit msg.

Cheers,
Michal

Reviewed-by on v1 1/1
#6 lore raw
In reply to Michal Orzel
Quoted 6 lines · Michal Orzel

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

Reported by clang-tidy [deadcode.DeadStores]

Quoted 24 lines
---
  ipc/sem.c | 2 --
  1 file changed, 2 deletions(-)
ipc/sem.c +0 -2
index 0dbdb98fdf2d..38ef91a63edd 100644
@@ -766,7 +766,6 @@ static int perform_atomic_semop(struct sem_array *sma, struct sem_queue *q)
  	for (sop = sops; sop < sops + nsops; sop++) {
  		curr = &sma->sems[sop->sem_num];
  		sem_op = sop->sem_op;
-		result = curr->semval;
 
  		if (sop->sem_flg & SEM_UNDO) {
  			int undo = un->semadj[sop->sem_num] - sem_op;
@@ -1430,7 +1429,6 @@ static int semctl_main(struct ipc_namespace *ns, int semid, int semnum,
  	if (err)
  		goto out_rcu_wakeup;
 
-	err = -EACCES;
  	switch (cmd) {
  	case GETALL:
  	{
Reviewed-by on v1 1/1
#7 lore raw
In reply to Michal Orzel

Hi Michal,

Quoted 6 lines · Michal Orzel

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

Reported by clang-tidy [deadcode.DeadStores]

Thank you for the patch! This should go through Andrew Morton, now CC'd.

For the future, having some history behind the origin of these warnings would make review a little easier. I'll comment on each below.

ipc/sem.c static int perform_atomic_semop(struct sem_array *sma, struct sem_queue *q) · line 769

Quoted 13 lines
---
 ipc/sem.c | 2 --
 1 file changed, 2 deletions(-)
ipc/sem.c +0 -1
index 0dbdb98fdf2d..38ef91a63edd 100644
@@ -766,7 +766,6 @@ static int perform_atomic_semop(struct sem_array *sma, struct sem_queue *q)
 	for (sop = sops; sop < sops + nsops; sop++) {
 		curr = &sma->sems[sop->sem_num];
 		sem_op = sop->sem_op;
-		result = curr->semval;

Caused by commit 4ce33ec2e42d ("ipc/sem: optimize perform_atomic_semop()"). Seems like a simple copy and paste bug.

ipc/sem.c static int semctl_main(struct ipc_namespace *ns, int semid, int semnum, · line 1433

Quoted 7 lines
ipc/sem.c +0 -1
 		if (sop->sem_flg & SEM_UNDO) {
 			int undo = un->semadj[sop->sem_num] - sem_op;
@@ -1430,7 +1429,6 @@ static int semctl_main(struct ipc_namespace *ns, int semid, int semnum,
 	if (err)
 		goto out_rcu_wakeup;
 
-	err = -EACCES;

I was not able to figure out this one but it does look correct, as err is either set properly in the switch statement below or we skip over it to the 'err = -EINVAL'.

Quoted 6 lines
ipc/sem.c +0 -0
 	switch (cmd) {
 	case GETALL:
 	{
Signature · 1 line
2.25.1
#8 lore raw

Hi Nathan,

Quoted 8 lines · Nathan Chancellor

Hi Michal,

Quoted 5 lines · Michal Orzel

Get rid of redundant assignments which end up in values not being read either because they are overwritten or the function ends.

Reported by clang-tidy [deadcode.DeadStores]

Quoted 9 lines
Quoted 1 line

Thank you for the patch! This should go through Andrew Morton, now CC'd.

For the future, having some history behind the origin of these warnings would make review a little easier. I'll comment on each below.

Thanks,
I will keep that in mind when sending other patches fixing static analyzer findings.

Cheers,
Michal