Bug 3277
| Summary: | raid0.c functions raid0_run() line 317: access invalid address | ||
|---|---|---|---|
| Product: | IO/Storage | Reporter: | raid (naww) |
| Component: | MD | Assignee: | Neil Brown (neilb) |
| Status: | RESOLVED PATCH_ALREADY_AVAILABLE | ||
| Severity: | normal | CC: | protasnb |
| Priority: | P2 | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Kernel Version: | 2.6.8.1 | Subsystem: | |
| Regression: | --- | Bisected commit-id: | |
| Attachments: | Patch that will be submitted upstream | ||
How is it working for you these days? Any better with new kernels? Thanks. The first problem was fixed by making the enclosing loop iterate once less often (it starts i=1 instead of i=0). The second is error is still present in the code but as the code is in an #ifdef that is never selected, it isn't a problem in practice. Nevertheless I will submit a patch to fix it. Created attachment 12921 [details]
Patch that will be submitted upstream
|
Distribution: Hardware Environment: Software Environment: Problem Description: raid0.c:functions raid0_run() line: 317 error: when cur > conf->nr_strip_zones (first zone is large, next some zones is small, so conf->hash_spaing is larger than the sum of next small zones), it would access an invalid address; fixed: while(size <= conf->hash_spacing ) --> while(size <= conf->hash_spacing && cur < conf->nr_strip_zones -1) type error: functions raid0_status() ,line 476: lost word seq ; fixed: seq_printf("(h%d)",h++); ---> seq_printf(seq,"(h%d)",h++); Steps to reproduce: