helper: accurate failure diagnostics + coherent library version#17
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: ASSERTIVE Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
📜 Recent review details⏰ Context from checks skipped due to timeout. (4)
🧰 Additional context used📓 Path-based instructions (1)meson.build📄 CodeRabbit inference engine (AGENTS.md)
Files:
🔇 Additional comments (2)
📝 WalkthroughSummary by CodeRabbit
WalkthroughThis PR updates helper error handling in both C copies, bumps the project and public version constants to 0.4.3, and changes the C and Rust version tests to compute the expected packed version from the macros. ChangesHelper diagnostics and version sync
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@bindings/rust/libdrmtap-sys/csrc/drmtap-helper.c`:
- Around line 191-200: The `drmtap-helper` error handling in the `cap_set_proc`
path has the same `errno`-loss issue as the other helper: `cap_free(caps)` can
change `errno` before `perror()` runs, causing misleading diagnostics. Update
the embedded helper logic around `cap_set_proc` in `drmtap-helper.c` to preserve
`errno` before calling `cap_free(caps)`, then restore it before the
`perror("drmtap-helper: cap_set_proc failed")` call so the reported error
remains accurate.
In `@bindings/rust/libdrmtap-sys/src/lib.rs`:
- Around line 149-151: The sys version test is still hardcoding the packed
version literal, so it can drift from the C library version. Update the
assertion in libdrmtap-sys/src/lib.rs to derive the expected value from the
exposed DRMTAP_VERSION_* constants instead of using 0x000404. Keep the check
near drmtap_version() and make it mirror how the C tests compute the packed
version so this stays in sync across releases.
In `@helper/drmtap-helper.c`:
- Around line 191-200: The error reporting in drmtap-helper around cap_set_flag
and cap_set_proc can lose the original failure cause because cap_free(caps) may
overwrite errno before perror() runs. In drmtap_helper.c, preserve errno
immediately after the failing libcap call, perform cleanup, then restore or use
the saved value when calling perror() in these error paths so the logged message
reflects the real failure.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 793e616d-fd64-4a4e-bd88-6a049108090c
📒 Files selected for processing (8)
bindings/rust/libdrmtap-sys/csrc/drmtap-helper.cbindings/rust/libdrmtap-sys/csrc/drmtap.hbindings/rust/libdrmtap-sys/src/lib.rshelper/drmtap-helper.cinclude/drmtap.hmeson.buildtests/test_enumerate.ctests/test_helper.c
📜 Review details
🧰 Additional context used
📓 Path-based instructions (6)
**/*.{c,h}
📄 CodeRabbit inference engine (AGENTS.md)
**/*.{c,h}: Use 4-space indentation, never tabs, and follow the documented C style conventions.
Usesnake_casefor functions and variables, prefix public API functions withdrmtap_, useUPPER_SNAKE_CASEfor macros/constants withDRMTAP_prefixes, and usesnake_case_tfor internal types anddrmtap_*for public types.
Use same-line braces (1TBS style).
Keep line length to a 100-character soft limit and 120-character hard limit.
Use//for single-line comments and/* */for multi-line comments.
Return negative errno values for errors, never abort, and always clean up resources on error paths.
Always checkmallocreturns and free allocated memory in cleanup paths.
Every.cand.hfile must start with the standard libdrmtap copyright/SPDX header block and an@file/@briefDoxygen file header.
Files:
tests/test_helper.ctests/test_enumerate.cinclude/drmtap.hbindings/rust/libdrmtap-sys/csrc/drmtap.hhelper/drmtap-helper.cbindings/rust/libdrmtap-sys/csrc/drmtap-helper.c
tests/*.c
📄 CodeRabbit inference engine (AGENTS.md)
tests/*.c: Write tests as simple assert-based C programs without an external test framework.
UseDRM_DEVICEfrom the environment in tests and do not hardcode/dev/dri/card0.
Do not test tiling/deswizzle behavior against vkms, because vkms is LINEAR-only.
Account for the fact that bare vkms has no active planes and may require a compositor for integration tests.
Files:
tests/test_helper.ctests/test_enumerate.c
**/*.h
📄 CodeRabbit inference engine (AGENTS.md)
Use
#ifndef DRMTAP_MODULE_H/#define DRMTAP_MODULE_Hstyle header guards.
Files:
include/drmtap.hbindings/rust/libdrmtap-sys/csrc/drmtap.h
include/drmtap.h
📄 CodeRabbit inference engine (AGENTS.md)
Public API functions in
include/drmtap.hmust have Doxygen comments documenting behavior, parameters, return values, and relevant error codes.
Files:
include/drmtap.h
meson.build
📄 CodeRabbit inference engine (AGENTS.md)
Use Meson as the build system configuration for the project.
Files:
meson.build
helper/*.c
📄 CodeRabbit inference engine (AGENTS.md)
The privileged helper binary should implement the helper spawn and SCM_RIGHTS IPC path.
Files:
helper/drmtap-helper.c
🔇 Additional comments (7)
helper/drmtap-helper.c (1)
212-258: LGTM!Also applies to: 273-287
bindings/rust/libdrmtap-sys/csrc/drmtap-helper.c (1)
212-258: LGTM!Also applies to: 273-287
meson.build (1)
3-4: LGTM!include/drmtap.h (1)
28-34: LGTM!bindings/rust/libdrmtap-sys/csrc/drmtap.h (1)
28-34: LGTM!tests/test_enumerate.c (1)
31-34: LGTM!tests/test_helper.c (1)
37-40: LGTM!
fxd0h
force-pushed
the
diag/helper-perror-version
branch
from
1f9ad68 to
afccf28
Compare
fxd0h
changed the title
|
Thanks, all three are addressed in the latest push.
|
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@bindings/rust/libdrmtap-sys/csrc/drmtap-helper.c`:
- Around line 191-198: The new single-line comments in drmtap-helper should use
C++-style line comments instead of block-comment syntax. Update the affected
comment sites in the helper flow around cap_set_flag, cap_set_proc, and the
other referenced spots so any added single-line notes are written with //, while
keeping any true multi-line explanations in /* */ form.
- Around line 286-290: The errno-backed grab failure paths are losing the saved
errno because send_error() only logs the reason string. Update the MODE_MAP_DUMB
failed and mmap failed call sites in drmtap-helper.c to preserve errno by
capturing it before any further calls, then either pass it into a helper or
format it alongside the reason; keep send_error() unchanged for the non-errno
callers since they do not need errno details.
In `@helper/drmtap-helper.c`:
- Around line 191-198: The added single-line comments in drmtap-helper.c are
using block-comment syntax instead of the required // style. Update the new
comments in the affected sections around cap_free(), cap_set_proc(), and the
other marked spots to use // for single-line notes, keeping /* */ only where a
true multi-line comment is needed.
- Around line 286-290: The current error reporting in send_error() only sends a
text reason, so the MODE_MAP_DUMB and mmap() failure paths lose the real
system-call failure cause. Update the call sites that report these failures to
capture errno immediately after the failing call and pass it through in the
diagnostic, or add a dedicated send_errno_error() helper that preserves and
reports errno alongside the reason. Keep the fix localized around send_error(),
the MODE_MAP_DUMB path, and the mmap() error handling so the original failure
details are not overwritten.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: f9ec6b23-ab23-44ba-aa7d-82cc9702f8f7
📒 Files selected for processing (8)
bindings/rust/libdrmtap-sys/csrc/drmtap-helper.cbindings/rust/libdrmtap-sys/csrc/drmtap.hbindings/rust/libdrmtap-sys/src/lib.rshelper/drmtap-helper.cinclude/drmtap.hmeson.buildtests/test_enumerate.ctests/test_helper.c
📜 Review details
🧰 Additional context used
📓 Path-based instructions (6)
**/*.{c,h}
📄 CodeRabbit inference engine (AGENTS.md)
**/*.{c,h}: Use 4-space indentation, never tabs, and follow the documented C style conventions.
Usesnake_casefor functions and variables, prefix public API functions withdrmtap_, useUPPER_SNAKE_CASEfor macros/constants withDRMTAP_prefixes, and usesnake_case_tfor internal types anddrmtap_*for public types.
Use same-line braces (1TBS style).
Keep line length to a 100-character soft limit and 120-character hard limit.
Use//for single-line comments and/* */for multi-line comments.
Return negative errno values for errors, never abort, and always clean up resources on error paths.
Always checkmallocreturns and free allocated memory in cleanup paths.
Every.cand.hfile must start with the standard libdrmtap copyright/SPDX header block and an@file/@briefDoxygen file header.
Files:
tests/test_helper.ctests/test_enumerate.cbindings/rust/libdrmtap-sys/csrc/drmtap.hinclude/drmtap.hbindings/rust/libdrmtap-sys/csrc/drmtap-helper.chelper/drmtap-helper.c
tests/*.c
📄 CodeRabbit inference engine (AGENTS.md)
tests/*.c: Write tests as simple assert-based C programs without an external test framework.
UseDRM_DEVICEfrom the environment in tests and do not hardcode/dev/dri/card0.
Do not test tiling/deswizzle behavior against vkms, because vkms is LINEAR-only.
Account for the fact that bare vkms has no active planes and may require a compositor for integration tests.
Files:
tests/test_helper.ctests/test_enumerate.c
meson.build
📄 CodeRabbit inference engine (AGENTS.md)
Use Meson as the build system configuration for the project.
Files:
meson.build
**/*.h
📄 CodeRabbit inference engine (AGENTS.md)
Use
#ifndef DRMTAP_MODULE_H/#define DRMTAP_MODULE_Hstyle header guards.
Files:
bindings/rust/libdrmtap-sys/csrc/drmtap.hinclude/drmtap.h
include/drmtap.h
📄 CodeRabbit inference engine (AGENTS.md)
Public API functions in
include/drmtap.hmust have Doxygen comments documenting behavior, parameters, return values, and relevant error codes.
Files:
include/drmtap.h
helper/*.c
📄 CodeRabbit inference engine (AGENTS.md)
The privileged helper binary should implement the helper spawn and SCM_RIGHTS IPC path.
Files:
helper/drmtap-helper.c
🔇 Additional comments (6)
meson.build (1)
4-4: LGTM!include/drmtap.h (1)
28-34: LGTM!bindings/rust/libdrmtap-sys/csrc/drmtap.h (1)
28-34: LGTM!tests/test_enumerate.c (1)
31-34: LGTM!tests/test_helper.c (1)
37-40: LGTM!bindings/rust/libdrmtap-sys/src/lib.rs (1)
149-157: LGTM!
| int saved = errno; /* cap_free() may clobber errno */ | ||
| cap_free(caps); | ||
| perror("cap_set_proc failed"); return -1; | ||
| errno = saved; | ||
| perror("drmtap-helper: cap_set_flag failed"); return -1; | ||
| } | ||
|
|
||
| int ret = cap_set_proc(caps); | ||
| int saved = errno; /* preserve before cap_free() touches errno */ |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win
Use // for added single-line comments.
These new single-line comments use block-comment syntax. As per coding guidelines, use // for single-line comments and /* */ for multi-line comments.
Also applies to: 216-217, 242-242, 256-256, 289-289
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@bindings/rust/libdrmtap-sys/csrc/drmtap-helper.c` around lines 191 - 198, The
new single-line comments in drmtap-helper should use C++-style line comments
instead of block-comment syntax. Update the affected comment sites in the helper
flow around cap_set_flag, cap_set_proc, and the other referenced spots so any
added single-line notes are written with //, while keeping any true multi-line
explanations in /* */ form.
Source: Coding guidelines
| int saved = errno; /* cap_free() may clobber errno */ | ||
| cap_free(caps); | ||
| perror("cap_set_proc failed"); return -1; | ||
| errno = saved; | ||
| perror("drmtap-helper: cap_set_flag failed"); return -1; | ||
| } | ||
|
|
||
| int ret = cap_set_proc(caps); | ||
| int saved = errno; /* preserve before cap_free() touches errno */ |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win
Use // for added single-line comments.
These new single-line comments use block-comment syntax. As per coding guidelines, use // for single-line comments and /* */ for multi-line comments.
Also applies to: 216-217, 242-242, 256-256, 289-289
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@helper/drmtap-helper.c` around lines 191 - 198, The added single-line
comments in drmtap-helper.c are using block-comment syntax instead of the
required // style. Update the new comments in the affected sections around
cap_free(), cap_set_proc(), and the other marked spots to use // for single-line
notes, keeping /* */ only where a true multi-line comment is needed.
Source: Coding guidelines
The privileged helper printed perror("cap_set_proc failed") on six error
paths, four of which had nothing to do with capabilities (cap_init,
cap_set_flag, seccomp_init, seccomp_rule_add, send, and the generic grab
error path). In a setcap binary a wrong error message actively misleads an
operator during incident response, so give each path an accurate message,
chosen by whether the failing call actually sets errno:
- cap_init / cap_set_flag / cap_set_proc / send: perror with the real call.
On the cap paths errno is saved before cap_free() and restored before
perror(), since the cleanup call may clobber it.
- seccomp_init: plain fprintf, it does not set errno so perror would print
a stale value.
- seccomp_rule_add / seccomp_load: they return -errno, so report
strerror(-rc) instead of perror.
The genuine cap_set_proc and seccomp_load failure paths now log instead of
returning silently.
The grab error path is split in two: send_error() for logical/validation
failures (no useful errno), and send_error_errno() for syscall/libdrm
failures (MODE_MAP_DUMB, mmap, drmModeGetFB2, drmModeGetPlaneResources),
which captures errno before send_all() can overwrite it and appends
strerror(). Previously every grab error printed a bogus "cap_set_proc
failed" via perror; this keeps the real cause where errno carries one and
drops the misleading string everywhere else.
drmtap_version() still reported 0.1.0 while the project is well past that,
which reads as unmaintained. Set DRMTAP_VERSION_* and the meson project
version to 0.4.3 so the C library reports the same version as the
libdrmtap-sys crate (same C sources). The version tests now derive their
expected value instead of hardcoding it: the C tests from DRMTAP_VERSION_*,
and the Rust sys test from CARGO_PKG_VERSION, which also cross-checks that
the C header and the crate version stay equal.
Closes #5
fxd0h
force-pushed
the
diag/helper-perror-version
branch
from
afccf28 to
94ff6ea
Compare
|
Round 2:
Re-validated locally: release+werror, asan+ubsan, plain/FORTIFY, meson test 5/5, cppcheck clean, cargo test green, real V3 capture OK. |
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@helper/drmtap-helper.c`:
- Around line 185-205: The capability and seccomp error paths in drmtap-helper
are still collapsing detailed failure codes into a generic -1. Update the
existing failure returns in the cap_init/cap_set_flag/cap_set_proc flow to
return the preserved negative errno value or the saved error code instead of -1,
and keep the cleanup/error logging intact. Also apply the same return-value
change to the seccomp_* paths so they propagate their existing rc/ret values,
and mirror the exact behavior in the duplicate drmtap-helper implementation
under the Rust bindings csrc.
In `@meson.build`:
- Line 4: Meson’s project version is still not tied to the exported/runtime
version values, so it can drift from DRMTAP_VERSION_* and CARGO_PKG_VERSION_*.
Update the version handling around meson.project_version() so the same source
drives the generated version macros, or add a configure/test-time assertion that
the Meson project version matches the library version symbols used by
drmtap_version(), tests/test_enumerate, tests/test_helper, and the Rust
bindings.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 9a75444c-32a0-428e-94a1-c6d965318771
📒 Files selected for processing (8)
bindings/rust/libdrmtap-sys/csrc/drmtap-helper.cbindings/rust/libdrmtap-sys/csrc/drmtap.hbindings/rust/libdrmtap-sys/src/lib.rshelper/drmtap-helper.cinclude/drmtap.hmeson.buildtests/test_enumerate.ctests/test_helper.c
📜 Review details
🧰 Additional context used
📓 Path-based instructions (6)
**/*.{c,h}
📄 CodeRabbit inference engine (AGENTS.md)
**/*.{c,h}: Use 4-space indentation, never tabs, and follow the documented C style conventions.
Usesnake_casefor functions and variables, prefix public API functions withdrmtap_, useUPPER_SNAKE_CASEfor macros/constants withDRMTAP_prefixes, and usesnake_case_tfor internal types anddrmtap_*for public types.
Use same-line braces (1TBS style).
Keep line length to a 100-character soft limit and 120-character hard limit.
Use//for single-line comments and/* */for multi-line comments.
Return negative errno values for errors, never abort, and always clean up resources on error paths.
Always checkmallocreturns and free allocated memory in cleanup paths.
Every.cand.hfile must start with the standard libdrmtap copyright/SPDX header block and an@file/@briefDoxygen file header.
Files:
tests/test_helper.cinclude/drmtap.htests/test_enumerate.cbindings/rust/libdrmtap-sys/csrc/drmtap.hhelper/drmtap-helper.cbindings/rust/libdrmtap-sys/csrc/drmtap-helper.c
tests/*.c
📄 CodeRabbit inference engine (AGENTS.md)
tests/*.c: Write tests as simple assert-based C programs without an external test framework.
UseDRM_DEVICEfrom the environment in tests and do not hardcode/dev/dri/card0.
Do not test tiling/deswizzle behavior against vkms, because vkms is LINEAR-only.
Account for the fact that bare vkms has no active planes and may require a compositor for integration tests.
Files:
tests/test_helper.ctests/test_enumerate.c
**/*.h
📄 CodeRabbit inference engine (AGENTS.md)
Use
#ifndef DRMTAP_MODULE_H/#define DRMTAP_MODULE_Hstyle header guards.
Files:
include/drmtap.hbindings/rust/libdrmtap-sys/csrc/drmtap.h
include/drmtap.h
📄 CodeRabbit inference engine (AGENTS.md)
Public API functions in
include/drmtap.hmust have Doxygen comments documenting behavior, parameters, return values, and relevant error codes.
Files:
include/drmtap.h
meson.build
📄 CodeRabbit inference engine (AGENTS.md)
Use Meson as the build system configuration for the project.
Files:
meson.build
helper/*.c
📄 CodeRabbit inference engine (AGENTS.md)
The privileged helper binary should implement the helper spawn and SCM_RIGHTS IPC path.
Files:
helper/drmtap-helper.c
🔇 Additional comments (2)
bindings/rust/libdrmtap-sys/csrc/drmtap-helper.c (1)
191-191: Use//for these single-line comments.Same guideline issue as the earlier review: these one-line notes are still written as
/* ... */instead of//.Also applies to: 198-198, 216-216, 242-242, 256-256, 288-288, 298-298
helper/drmtap-helper.c (1)
191-191: Use//for these single-line comments.Same guideline issue as the earlier review: these one-line notes are still written as
/* ... */instead of//.Also applies to: 198-198, 216-216, 242-242, 256-256, 288-288, 298-298
| perror("drmtap-helper: cap_init failed"); return -1; | ||
| } | ||
|
|
||
| cap_value_t keep[] = { CAP_SYS_ADMIN }; | ||
| if (cap_set_flag(caps, CAP_PERMITTED, 1, keep, CAP_SET) != 0 || | ||
| cap_set_flag(caps, CAP_EFFECTIVE, 1, keep, CAP_SET) != 0) { | ||
| int saved = errno; /* cap_free() may clobber errno */ | ||
| cap_free(caps); | ||
| perror("cap_set_proc failed"); return -1; | ||
| errno = saved; | ||
| perror("drmtap-helper: cap_set_flag failed"); return -1; | ||
| } | ||
|
|
||
| int ret = cap_set_proc(caps); | ||
| int saved = errno; /* preserve before cap_free() touches errno */ | ||
| cap_free(caps); | ||
|
|
||
| if (ret == 0) { | ||
| fprintf(stderr, "drmtap-helper: dropped caps, keeping CAP_SYS_ADMIN\n"); | ||
| if (ret != 0) { | ||
| errno = saved; | ||
| perror("drmtap-helper: cap_set_proc failed"); | ||
| return -1; | ||
| } |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟠 Major | ⚡ Quick win
Return the specific error codes you already have.
These new paths now log the right failure, but they still collapse errno / libseccomp's -errno into -1. That drops machine-readable detail and conflicts with the C guideline to return negative errno values. Please return -errno/-saved for the capability failures and rc/ret for the seccomp_* failures, then mirror the same adjustment into bindings/rust/libdrmtap-sys/csrc/drmtap-helper.c.
Suggested shape
- perror("drmtap-helper: cap_init failed"); return -1;
+ int saved = errno;
+ perror("drmtap-helper: cap_init failed");
+ return saved ? -saved : -1;
...
- perror("drmtap-helper: cap_set_flag failed"); return -1;
+ perror("drmtap-helper: cap_set_flag failed");
+ return saved ? -saved : -1;
...
- perror("drmtap-helper: cap_set_proc failed");
- return -1;
+ perror("drmtap-helper: cap_set_proc failed");
+ return saved ? -saved : -1;
...
- return -1;
+ return rc;
...
- return -1;
+ return ret;Also applies to: 242-259
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@helper/drmtap-helper.c` around lines 185 - 205, The capability and seccomp
error paths in drmtap-helper are still collapsing detailed failure codes into a
generic -1. Update the existing failure returns in the
cap_init/cap_set_flag/cap_set_proc flow to return the preserved negative errno
value or the saved error code instead of -1, and keep the cleanup/error logging
intact. Also apply the same return-value change to the seccomp_* paths so they
propagate their existing rc/ret values, and mirror the exact behavior in the
duplicate drmtap-helper implementation under the Rust bindings csrc.
Source: Coding guidelines
The library version lives in the meson project() version, the DRMTAP_VERSION_* macros in include/drmtap.h, and the Rust crate version. The version unit tests already tie the header macros to drmtap_version() and to CARGO_PKG_VERSION; this closes the last gap by parsing the public header at configure time and asserting it equals meson.project_version(), so a future bump that touches one source but not the other fails the build early with a clear message instead of drifting silently.
|
Round 3:
Re-validated: release/asan/plain reconfigure clean, meson test 5/5. |
1 participant
Addresses the two code-rigor findings in #5 from the RustDesk maintainer review.
Misleading diagnostics
The helper printed
perror("cap_set_proc failed")on six error paths, four of which have nothing to do with capabilities. In a setcap binary a wrong message actively misleads an operator during incident response (a seccomp or socket failure reporting a capability failure). Each path now gets an accurate message, chosen by whether the failing call actually setserrno:cap_initperror(cap_init failed)cap_set_flagperror(cap_set_flag failed)seccomp_initfprintf(seccomp_init failed)(does not set errno, so no perror)seccomp_rule_addfprintf(... strerror(-rc))(returns -errno, does not set errno)sendperror(send failed)send_errorOn the two cap paths,
errnois now saved beforecap_free()and restored beforeperror(), since the cleanup call may clobber it (would have reintroduced the same class of misleading message). The genuinecap_set_procandseccomp_loadfailure paths now log instead of returning silently.Coherent version
drmtap_version()returned 0.1.0 while the project is well past that, which reads as unmaintained. SetDRMTAP_VERSION_*and the meson project version to 0.4.3, matching the currently published libdrmtap-sys crate (same C sources). The.sosoversion stays 0 (ABI unchanged); only the full version string moves. The version tests now derive their expected value instead of hardcoding it:test_helper,test_enumerate): fromDRMTAP_VERSION_*.CARGO_PKG_VERSION, which also cross-checks that the C header and the crate version stay equal (fails loudly if a future release bumps one without the other).Validation (local, x86_64 i915)
Closes #5