On 12/6/17 5:16 AM, Peter Zijlstra wrote:
On Wed, Dec 06, 2017 at 12:56:36PM +0100, Peter Zijlstra wrote:
On Tue, Dec 05, 2017 at 10:31:28PM -0800, Yonghong Song wrote:
Commit e87c6bc3852b ("bpf: permit multiple bpf attachments
for a single perf event") added support to attach multiple
bpf programs to a single perf event.
Commit 2541517c32be ("tracing, perf: Implement BPF programs
attached to kprobes") utilized the existing perf ioctl
interface and added the command PERF_EVENT_IOC_SET_BPF
to attach a bpf program to a tracepoint.
This patch adds a new ioctl
command, given a perf event fd, to query the bpf program array
attached to the same perf tracepoint event.
The new uapi ioctl command:
PERF_EVENT_IOC_QUERY_BPF
The new uapi/linux/perf_event.h structure:
struct perf_event_query_bpf {
__u64 prog_ids;
__u32 prog_cnt;
};
The usage:
struct perf_event_query_bpf query;
query.prog_ids = (__u64)usr_prog_ids_buf;
query.prog_cnt = usr_prog_ids_buf_len;
err = ioctl(pmu_efd, PERF_EVENT_IOC_QUERY_BPF, &query);
Signed-off-by: Yonghong Song <y...@fb.com>
Acked-by: Alexei Starovoitov <a...@kernel.org>
Can you please fix that example to make it clear that prog_ids is in
fact a pointer to an array of size prog_cnt. Ideally also describing
what the type of array is.
Right. Will address this with more descriptions in the commit message
and also add some comments in the perf_event.h.
In fact, would not something like:
struct perf_event_query_bpf {
__u32 len;
__u32 __reserved;
I suppose we could use this field to store the number of entries
returned, retaining the len to indicate how large the structure is.
__u64 ids[0];
};
be a much clearer interface?
Yes, this is clearer and may be consistent with perf interface.
FYI, my old interface is similar to the BPF query interface below:
struct { /* anonymous struct used by BPF_PROG_QUERY command */
__u32 target_fd; /* container object to
query */
__u32 attach_type;
__u32 query_flags;
__u32 attach_flags;
__aligned_u64 prog_ids;
__u32 prog_cnt;
} query;
Also, you forgot to tell us why we need this interface at all.
Right. Will add some descriptions for this.
