map_seq_next should increase position index
PATCH 1/1 map_seq_next should increase position index
if seq_file .next fuction does not change position index, read after some lseek can generate unexpected output.
https://bugzilla.kernel.org/show_bug.cgi?id=206283
--- kernel/bpf/inode.c | 1 + 1 file changed, 1 insertion(+)
kernel/bpf/inode.c +1 -0
@@ -196,6 +196,7 @@ static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos)void *key =map_iter (m)->key;void *prev_key; + (*pos)++;if (map_iter (m)->done)return NULL ;
Signature · 1 line
1.8.3.1
Daniel Borkmann
maintainerQuoted 15 lines · Vasily Averin
if
seq_file.next fuction does not change position index, read after some lseek can generate unexpected output.https://bugzilla.kernel.org/show_bug.cgi?id=206283
Signed-off-by Vasily Averin <[email protected]>--- kernel/bpf/inode.c | 1 + 1 file changed, 1 insertion(+)kernel/bpf/inode.c +0
-0
kernel/bpf/inode.c static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos) · line 202
Quoted 7 lines
kernel/bpf/inode.c +1
-0@@ -196,6 +196,7 @@ static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos)void *key =map_iter (m)->key;void *prev_key; + (*pos)++;if (map_iter (m)->done)return NULL ;
Hm, how did you test this change? Please elaborate, since in map_seq_next() we do increment position index:
static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos)
{
struct bpf_map *map = seq_file_to_map(m);
void *key = map_iter(m)->key;
void *prev_key;
if (map_iter(m)->done)
return NULL;
if (unlikely(v == SEQ_START_TOKEN))
prev_key = NULL;
else
prev_key = key;
if (map->ops->map_get_next_key(map, prev_key, key)) {
map_iter(m)->done = true;
return NULL;
}
++(*pos); <------------ here
return key;
}With your change we'd increment twice. What is missing here?
Thanks,
Daniel
Quoted 10 lines · Daniel Borkmann
Quoted 9 lines · Vasily Averin
if
seq_file.next fuction does not change position index, read after some lseek can generate unexpected output.https://bugzilla.kernel.org/show_bug.cgi?id=206283
Signed-off-by Vasily Averin <[email protected]>---
kernel/bpf/inode.c| 1 +1 file changed, 1 insertion(+)
Quoted 14 lines
Quoted 11 lines
kernel/bpf/inode.c +0
-0@@ -196,6 +196,7 @@ static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos) void *key = map_iter(m)->key; void *prev_key; + (*pos)++; if (map_iter(m)->done) return NULL;Hm, how did you test this change? Please elaborate, since in
map_seq_next()we do increment position index:
Position index should be updated even if .next returns NULL.
Sorry, I've not tested this case.
$ dd if=AFFECTED_FILE bs=1000 skip=1
With any huge bs it will generate last line.
If you'll specify bs in middle of last line --
dd will output rest of list line and then whole last line once again.
Quoted 14 lines
static void *map_seq_next(struct
seq_file*m, void *v,loff_t*pos) {
structbpf_map*map =seq_file_to_map(m);
void *key = map_iter(m)->key;
void*prev_key;if (map_iter(m)->done)
return NULL;if (
unlikely(v == SEQ_START_TOKEN))
prev_key= NULL;
else
prev_key= key;
Quoted 6 lines
if (
map->ops->map_get_next_key(map, prev_key, key)) { map_iter(m)->done = true;
return NULL;
}++(*pos); <------------ here
You are right, I've missed it, it should be removed.
Quoted 4 lines
return key;
}With your change we'd increment twice. What is missing here?
Daniel Borkmann
maintainerQuoted 50 lines · Vasily Averin
Quoted 10 lines · Daniel Borkmann
Quoted 9 lines · Vasily Averin
if
seq_file.next fuction does not change position index, read after some lseek can generate unexpected output.https://bugzilla.kernel.org/show_bug.cgi?id=206283
Signed-off-by Vasily Averin <[email protected]>--- kernel/bpf/inode.c | 1 + 1 file changed, 1 insertion(+)Quoted 14 lines
kernel/bpf/inode.c static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos) · line 201
Quoted 11 lines
kernel/bpf/inode.c +0
-0@@ -196,6 +196,7 @@ static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos) void *key = map_iter(m)->key; void *prev_key; + (*pos)++; if (map_iter(m)->done) return NULL;Hm, how did you test this change? Please elaborate, since in
map_seq_next()we do increment position index:Position index should be updated even if .next returns NULL.
Sorry, I've not tested this case.
$ dd if=AFFECTED_FILE bs=1000 skip=1With any huge bs it will generate last line.
If you'll specify bs in middle of last line --
dd will output rest of list line and then whole last line once again.Quoted 14 lines
static void *map_seq_next(struct
seq_file*m, void *v,loff_t*pos) {
structbpf_map*map =seq_file_to_map(m);
void *key = map_iter(m)->key;
void*prev_key;if (map_iter(m)->done)
return NULL;if (
unlikely(v == SEQ_START_TOKEN))
prev_key= NULL;
else
prev_key= key;
Quoted 8 lines
Quoted 6 lines
if (
map->ops->map_get_next_key(map, prev_key, key)) { map_iter(m)->done = true;
return NULL;
}++(*pos); <------------ here
You are right, I've missed it, it should be removed.
Ok, please respin v2 then.
Thanks,
Daniel
PATCH v2 1/1 map_seq_next should increase position index
v2: removed missed increment in end of function
if seq_file .next fuction does not change position index, read after some lseek can generate unexpected output.
https://bugzilla.kernel.org/show_bug.cgi?id=206283
--- kernel/bpf/inode.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
kernel/bpf/inode.c +1 -2
@@ -196,6 +196,7 @@ static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos)void *key =map_iter (m)->key;void *prev_key; + (*pos)++;if (map_iter (m)->done)return NULL ;
@@ -208,8 +209,6 @@ static void *map_seq_next(struct seq_file *m, void *v, loff_t *pos)map_iter (m)->done = true;return NULL ; } - - ++(*pos);return key; }
Signature · 1 line
1.8.3.1
Quoted 7 lines · Vasily Averin
v2: removed missed increment in end of function
if
seq_file.next fuction does not change position index, read after some lseek can generate unexpected output.https://bugzilla.kernel.org/show_bug.cgi?id=206283
Signed-off-by Vasily Averin <[email protected]>
Applied, thanks!