You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
What reproduces the bug? Provide code if possible.
When using globbing to match multiple kprobes we use the multi kprobe link type to attach the probes. When doing this we report the number of probes being attached to as 1 instead of the actual number of probes being attached.
Technically speaking, there's only one probe being attached (from bpftrace point of view) with the list of functions passed via link_opts. OTOH, I see that this may be confusing.
At the moment, we print the amount of attached functions, along with the complete list of functions, when -v is used:
I think it's important here not to introduce anything that could be confusing to an end user as it's confusing enough as it is for them. IMO there should be no difference in the output given to a user whether they specified the probes manually or whether they obtained the list of probes to enable via globbing. Exposing the details of a 1:N mapping that kprobe multi gives could be very confusing, especially if the output is different to the non-globbed way of specifying the probes.
What reproduces the bug? Provide code if possible.
When using globbing to match multiple kprobes we use the multi kprobe link type to attach the probes. When doing this we report the number of probes being attached to as 1 instead of the actual number of probes being attached.
Obviously we can calculate up front how many probes we are trying to attach to though:
I haven't tested but this problem probably applies to uprobe multi attach as well.
bpftrace --info
outputThe text was updated successfully, but these errors were encountered: