Daniel Xu

Deleting scalar map just zeros out the entry

2

### What reproduces the bug? Provide code if possible. ``` $ sudo ./build/src/bpftrace -e 'BEGIN { @=1; delete(@); exit() }' Attaching 1 probe... @: 0 ``` This was added in...

Drop as many capabilities as we can

2

### Is your feature request related to a problem? Please describe. CAP_PERFMON and CAP_BPF are landed in 5.8 kernel ( https://lwn.net/Articles/827820/ ). We should consider dropping CAP_SYS_ADMIN b/c less caps...

Move all script-dependent stack allocations to scratch maps

1

Obviously at the program level the aggregate stack usage is always script dependent. This issue is specific to **individual** stack allocations that are script dependent. This attempts to tackle the...

Kernel targets print extraneous shell prompt at end

1

``` $ ./target/debug/vmtest -k tests/.assets/bzImage-v6.2-empty ls => bzImage-v6.2-empty ===> Booting ===> Setting up VM ===> Running command Cargo.lock Cargo.toml docs LICENSE Makefile README.md scripts src target tests vmtest.toml [root@(none) /]#...

Migrate usdt attachment from bcc to libbpf API

13

`AttachedProbe::attach_usdt()` uses bcc's `bpf_attach_uprobe()` to attach usdt probes (which are actually just backed by uprobes). This can be migrated to libbpf's `bpf_program__attach_uprobe()`. Note that `attach_usdt()` makes use of a few...

Overhaul release assets

7

Our current release assets are a bit messy. Take for example https://github.com/bpftrace/bpftrace/releases/tag/v0.22.1. We upload: * binary_tools_man-bundle.tar.xz * binary_tools_man-bundle.tar.zst * bpftrace * man.tar.xz * tools-with-help.tar.xz * tools.tar.xz The intent was to...

Support map-in-map

2

### Is your feature request related to a problem? Please describe. One thing that's been requested is for bpftrace to leverage map-in-map. Kernel support map-in-map, where you can place multiple...

Check all known program loss counters

1

See commits for more details. TODO: write tests This closes #835. ##### Checklist - [ ] Language changes are updated in `man/adoc/bpftrace.adoc` - [ ] User-visible and non-trivial changes updated...

Split ProbeType::watchpoint into ProbeType::kwatchpoint and ProbeType::watchpoint

1

B/c address spaces can overlap on some architectures and users can do things that require probe_read()s in probe body: https://github.com/iovisor/bpftrace/blob/79d312d56c11b910d6f10f4917d3cbc5d9982bbc/src/ast/semantic_analyser.cpp#L204 Also see discussion in https://github.com/iovisor/bpftrace/pull/1078#issuecomment-573961619