[PATCH bpf] tools/bpftool: fix PID fetching with a lot of results

[PATCH bpf] tools/bpftool: fix PID fetching with a lot of results

[Andrii Nakryiko]

In case of having so many PID results that they don't fit into a singe page
(4096) bytes, bpftool will erroneously conclude that it got corrupted data due
to 4096 not being a multiple of struct pid_iter_entry, so the last entry will
be partially truncated. Fix this by sizing the buffer to fit exactly N entries
with no truncation in the middle of record.

Fixes: d53dee3fe013 ("tools/bpftool: Show info for processes holding BPF map/prog/link/btf FDs")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
---
 tools/bpf/bpftool/pids.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/bpf/bpftool/pids.c b/tools/bpf/bpftool/pids.c
index df7d8ec76036..477e55d59c34 100644
--- a/tools/bpf/bpftool/pids.c
+++ b/tools/bpf/bpftool/pids.c
@@ -89,9 +89,9 @@ libbpf_print_none(__maybe_unused enum libbpf_print_level level,
 
 int build_obj_refs_table(struct obj_refs_table *table, enum bpf_obj_type type)
 {
-	char buf[4096];
-	struct pid_iter_bpf *skel;
 	struct pid_iter_entry *e;
+	char buf[4096 / sizeof(*e) * sizeof(*e)];
+	struct pid_iter_bpf *skel;
 	int err, ret, fd = -1, i;
 	libbpf_print_fn_t default_print;
 
-- 
2.24.1

Re: [PATCH bpf] tools/bpftool: fix PID fetching with a lot of results

[Yonghong Song]

On 12/4/20 3:20 PM, Andrii Nakryiko wrote:
> In case of having so many PID results that they don't fit into a singe page
> (4096) bytes, bpftool will erroneously conclude that it got corrupted data due
> to 4096 not being a multiple of struct pid_iter_entry, so the last entry will
> be partially truncated. Fix this by sizing the buffer to fit exactly N entries
> with no truncation in the middle of record.
> 
> Fixes: d53dee3fe013 ("tools/bpftool: Show info for processes holding BPF map/prog/link/btf FDs")
> Signed-off-by: Andrii Nakryiko <andrii@kernel.org>

Ack with one nit below.

Acked-by: Yonghong Song <yhs@fb.com>

> ---
>   tools/bpf/bpftool/pids.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/tools/bpf/bpftool/pids.c b/tools/bpf/bpftool/pids.c
> index df7d8ec76036..477e55d59c34 100644
> --- a/tools/bpf/bpftool/pids.c
> +++ b/tools/bpf/bpftool/pids.c
> @@ -89,9 +89,9 @@ libbpf_print_none(__maybe_unused enum libbpf_print_level level,
>   
>   int build_obj_refs_table(struct obj_refs_table *table, enum bpf_obj_type type)
>   {
> -	char buf[4096];
> -	struct pid_iter_bpf *skel;
>   	struct pid_iter_entry *e;
> +	char buf[4096 / sizeof(*e) * sizeof(*e)];
> +	struct pid_iter_bpf *skel;

No need to move "struct pid_iter_bpf *skel", right?

>   	int err, ret, fd = -1, i;
>   	libbpf_print_fn_t default_print;
>   
> 

Re: [PATCH bpf] tools/bpftool: fix PID fetching with a lot of results

[Andrii Nakryiko]

On Sat, Dec 5, 2020 at 11:11 AM Yonghong Song <yhs@fb.com> wrote:
>
>
>
> On 12/4/20 3:20 PM, Andrii Nakryiko wrote:
> > In case of having so many PID results that they don't fit into a singe page
> > (4096) bytes, bpftool will erroneously conclude that it got corrupted data due
> > to 4096 not being a multiple of struct pid_iter_entry, so the last entry will
> > be partially truncated. Fix this by sizing the buffer to fit exactly N entries
> > with no truncation in the middle of record.
> >
> > Fixes: d53dee3fe013 ("tools/bpftool: Show info for processes holding BPF map/prog/link/btf FDs")
> > Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
>
> Ack with one nit below.
>
> Acked-by: Yonghong Song <yhs@fb.com>
>
> > ---
> >   tools/bpf/bpftool/pids.c | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/tools/bpf/bpftool/pids.c b/tools/bpf/bpftool/pids.c
> > index df7d8ec76036..477e55d59c34 100644
> > --- a/tools/bpf/bpftool/pids.c
> > +++ b/tools/bpf/bpftool/pids.c
> > @@ -89,9 +89,9 @@ libbpf_print_none(__maybe_unused enum libbpf_print_level level,
> >
> >   int build_obj_refs_table(struct obj_refs_table *table, enum bpf_obj_type type)
> >   {
> > -     char buf[4096];
> > -     struct pid_iter_bpf *skel;
> >       struct pid_iter_entry *e;
> > +     char buf[4096 / sizeof(*e) * sizeof(*e)];
> > +     struct pid_iter_bpf *skel;
>
> No need to move "struct pid_iter_bpf *skel", right?

It's actually a move of `struct pid_iter_entry *e;` in from of char
buf[], to be able to use sizeof(*e) instead of sizeof(struct
pid_iter_bpf). It's just that diff tool didn't catch this properly :)

>
> >       int err, ret, fd = -1, i;
> >       libbpf_print_fn_t default_print;
> >
> >

Re: [PATCH bpf] tools/bpftool: fix PID fetching with a lot of results

[Yonghong Song]

On 12/7/20 6:55 PM, Andrii Nakryiko wrote:
> On Sat, Dec 5, 2020 at 11:11 AM Yonghong Song <yhs@fb.com> wrote:
>>
>>
>>
>> On 12/4/20 3:20 PM, Andrii Nakryiko wrote:
>>> In case of having so many PID results that they don't fit into a singe page
>>> (4096) bytes, bpftool will erroneously conclude that it got corrupted data due
>>> to 4096 not being a multiple of struct pid_iter_entry, so the last entry will
>>> be partially truncated. Fix this by sizing the buffer to fit exactly N entries
>>> with no truncation in the middle of record.
>>>
>>> Fixes: d53dee3fe013 ("tools/bpftool: Show info for processes holding BPF map/prog/link/btf FDs")
>>> Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
>>
>> Ack with one nit below.
>>
>> Acked-by: Yonghong Song <yhs@fb.com>
>>
>>> ---
>>>    tools/bpf/bpftool/pids.c | 4 ++--
>>>    1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/tools/bpf/bpftool/pids.c b/tools/bpf/bpftool/pids.c
>>> index df7d8ec76036..477e55d59c34 100644
>>> --- a/tools/bpf/bpftool/pids.c
>>> +++ b/tools/bpf/bpftool/pids.c
>>> @@ -89,9 +89,9 @@ libbpf_print_none(__maybe_unused enum libbpf_print_level level,
>>>
>>>    int build_obj_refs_table(struct obj_refs_table *table, enum bpf_obj_type type)
>>>    {
>>> -     char buf[4096];
>>> -     struct pid_iter_bpf *skel;
>>>        struct pid_iter_entry *e;
>>> +     char buf[4096 / sizeof(*e) * sizeof(*e)];
>>> +     struct pid_iter_bpf *skel;
>>
>> No need to move "struct pid_iter_bpf *skel", right?
> 
> It's actually a move of `struct pid_iter_entry *e;` in from of char
> buf[], to be able to use sizeof(*e) instead of sizeof(struct
> pid_iter_bpf). It's just that diff tool didn't catch this properly :)

Indeed. Looking at the final code, no unnecessary code churn.

> 
>>
>>>        int err, ret, fd = -1, i;
>>>        libbpf_print_fn_t default_print;
>>>
>>>

Re: [PATCH bpf] tools/bpftool: fix PID fetching with a lot of results

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to bpf/bpf.git (refs/heads/master):

On Fri, 4 Dec 2020 15:20:01 -0800 you wrote:
> In case of having so many PID results that they don't fit into a singe page
> (4096) bytes, bpftool will erroneously conclude that it got corrupted data due
> to 4096 not being a multiple of struct pid_iter_entry, so the last entry will
> be partially truncated. Fix this by sizing the buffer to fit exactly N entries
> with no truncation in the middle of record.
> 
> Fixes: d53dee3fe013 ("tools/bpftool: Show info for processes holding BPF map/prog/link/btf FDs")
> Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
> 
> [...]

Here is the summary with links:
  - [bpf] tools/bpftool: fix PID fetching with a lot of results
    https://git.kernel.org/bpf/bpf/c/932c60558109

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html