| Summary: | [Qemux86-64] v86d segfault in libc-2.10.1.so | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] Meta-yocto | Reporter: | Jiajun Xu <jiajun.xu> |
| Component: | meta-yocto | Assignee: | Bruce Ashfield <bruce.ashfield> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium | CC: | edwin.zhai, jessica.zhang, josh, kevin.tian, kunwei.tian, poky.bs.watcher, poky.watcher, richard.purdie, sgw |
| Version: | 0.9 | ||
| Target Milestone: | 0.9 M4 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Program Management | Documentation change: | --- |
|
Description
Jiajun Xu
2010-09-06 19:51:45 UTC
The segfault comes from a memcpy, where the sending buffer of uvesafb is copied to receiving buffer of v86d. The root cause of the segfault is that these 2 buffers has different size: uvesafb is 65536, but v86d is 16384. The different size is because that we use linux-wrs kernels(decides the uvesafb buffer size), and linux-libc-headers(2.6.33, decides the v86d's). Although this dismatch, the heads should be same as they are seldom changed. But recently the buffer size is enlarged from 16384 to 65536 on wrs kernel (701c502be984c38e242beca0e8e69903d9ead6b4), which cause this problem. Bruce is working on changes to the linux-libc-headers such that we should have the correct connector size in the headers for qemux86-64 following his next patch queue submission. (In reply to comment #2) > Bruce is working on changes to the linux-libc-headers such that we should have > the correct connector size in the headers for qemux86-64 following his next > patch queue submission. Josh, Do you mean changing the linux-libc-headers-wrs? I have sent out pull request of syncing the linux-libc-headers with wrs kernels, just enlarge the size as same as wrs kernels. Thanks, edwin See my comments on the mailing list. Once I'm done refactoring the kernel recipe, the headers recipe will have its checkout fixed and that patch will always be present. So we shouldn't need to patch the kernel headers at all. Checked with 20101007 build, the bug is fixed. |