View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0002318 | Xdebug | Step Debugging | public | 2025-02-12 10:36 | 2025-02-18 11:51 |
| Reporter | savemetenminutes | Assigned To | derick | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | assigned | Resolution | open | ||
| Product Version | 3.4.1 | ||||
| Summary | 0002318: Another segfault report | ||||
| Description | This issue has been reported multiple times. I'm not sure how to label the reproducibility of the issue as in certain scenarios the issue is reproducible without failure, but with other scripts/breakpoints/run to cursor actions it never manifests. I've labeled it as always reproducible, as with my current test case and steps I'm undergoing it will crash and produce exactly the same step debug log every time. I'd say that the frequency of occurrence on my environment is around 40%-50% of all step debug sessions I go though during my daily routine. I haven't been able to reproduce it using simple "1 file" or "20-30 lines of code" scripts, so I'm afraid that I cannot fulfil one of the requirements of bug reporting as described in the Xdebug documentation. I went through the generated step debug log generated by xdebug.log=/tmp/xdebug.log, but I don't think that it contains the culprit of the issue. Otherwise the issue is pretty straightforward to explain. Using PhpStorm, after stopping at a cursor position using "Force run to cursor" or at a breakpoint, using "Force run to cursor" again will result in the debug session and script to terminate. I can confirm that this issue started manifesting after PHP version 8(.1?). As I've pointed out in a comment on the first referenced issue above:
Nginx: 2024/01/05 12:41:58 [error] 47#47: *10 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 223.223.0.1, server: $host, request: "GET /some-resource/1?XDEBUG_SESSION=asd HTTP/1.1", upstream: "fastcgi://223.223.0.37:9090", host: "127.0.0.1:19492" [Fri Jan 05 12:38:48.373555 2024] [proxy_fcgi:error] [pid 34:tid 140142342183736] [client 223.223.0.1:52312] AH01067: Failed to read FastCGI header | ||||
| Steps To Reproduce | PHP 8.1 | ||||
| Additional Information | Somewhere I came across a suggestion to enable "develop" in the xdebug.mode setting (to enable the built-in stack overflow prevention? IIRC). This caused my debug session to hang forever. It also has the unwanted effect of modifying the var_dump functions and the exceptions stringification becomes HTML based if I remember correctly. | ||||
| Tags | crash, segfault | ||||
| Operating System | |||||
| PHP Version | 8.1.30-8.1.39 | ||||
|
|
I cannot find an edit button to fix my quote in the description. I'm also having trouble uploading/attaching files. |
|
|
Then how can I? If I can't reproduce it, I can not fix it. |
|
|
I would think that there would be comprehensive enough logs and diagnostic tools to be able to catch such sort of bugs similar to how an APM works in a regular PHP project. On our production environments we do not require customers to provide us with screenshots and browser request/response dumps. We monitor the Elastic logs and catch the exceptions before they are reported and we can obtain the request data and stack traces of where the code errors out. I'm actually kind of surprised that the mindset is to put away such a prevalent issue, which has been lingering and reported a multitude of times not only on this issue tracker and label it as unreproducible. That aside, I can reproduce it at will. And I'm willing to install whatever tracer/debug tool you might have on your end and contribute actively towards diagnosing and providing whatever you need as info or even provide live feed on the subject or multple screen recodings if that will help. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2025-02-12 10:36 | savemetenminutes | New Issue | |
| 2025-02-12 10:36 | savemetenminutes | Tag Attached: crash | |
| 2025-02-12 10:36 | savemetenminutes | Tag Attached: segfault | |
| 2025-02-12 10:42 | savemetenminutes | Note Added: 0007173 | |
| 2025-02-12 10:42 | savemetenminutes | File Added: xdebug.log | |
| 2025-02-12 14:30 | derick | Assigned To | => derick |
| 2025-02-12 14:30 | derick | Status | new => feedback |
| 2025-02-12 14:30 | derick | Note Added: 0007174 | |
| 2025-02-13 11:56 | savemetenminutes | Note Added: 0007175 | |
| 2025-02-13 11:56 | savemetenminutes | Status | feedback => assigned |
| 2025-02-18 11:51 | derick | View Status | private => public |
