Log errors in newFrameMapping as promised

[bobrik]

In synchronizeMappings there's this comment:

			fm, err = pm.newFrameMapping(pr, m)
			if err != nil {
				// newFrameMapping logged the message if needed
				continue
			}

With this change we actually do log errors, which can cause failure to unwind.

---

In #1230 I noticed continued failures to unwind. I was able to find some unlogged errors that explain things.

When unwinding fails, I can see a sea of these in strace:

[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092428] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092428] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092428] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092428] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092428] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092428] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092455] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0
[pid 2092513] newfstatat(AT_FDCWD, "/usr/local/bin/fl2", {st_mode=S_IFREG|0755, st_size=1088178912, ...}, 0) = 0

The error from delve:

> go.opentelemetry.io/ebpf-profiler/processmanager.(*ProcessManager).newFrameMapping() /home/gitlab-runner/go/pkg/mod/github.com/bobrik/opentelemetry-ebpf-profiler@v0.0.0-20260323182331-4ea65558c19d/processmanager/processinfo.go:359 (PC: 0xd190b6)
Warning: debugging optimized function
(dlv) print err
error(*fmt.wrapError) *{
        msg: "failed to load deltas: failed UpdateStackDeltaPages for FileID 2...+74 more",
        err: error(*errors.errorString) *{
                s: "failed UpdateStackDeltaPages for FileID 24c958ecb917576b: batch ...+51 more",},}

There's an overflowing map:

ivan@11m700:~$ MAP_ID=$(sudo bpftool map show -p | jq -r '.[] | select(.type == "hash") | select(.name != null) | select(.name == "stack_delta_pag") | .id')
ivan@11m700:~$ sudo bpftool map dump id $MAP_ID 2>/dev/null | grep -c '"key":'
65536

It does not overflow on a good machine that can unwind (barely):

ivan@11m672:~$ sudo bpftool map dump id $MAP_ID 2>/dev/null | grep -c '"key":'
64962

The fix for me is to increase the map scale factor, but we should also log the errors to make this failure more obvious.

[fabled]

lgtm