By default, Bugzilla does not search the list of RESOLVED bugs.You can force it to do so by putting the upper-case word ALL in front of your search query, e.g.: ALL tdelibs
We recommend searching for bugs this way, as you may discover that your bug has already been resolved and fixed in a later release.
| Summary: |
libr0 bug causes FTBFS when build is in different filesystem than /tmp |
| Product: |
TDE
|
Reporter: |
mgb-trinity |
| Component: |
non-core programs | Assignee: |
Timothy Pearson <kb9vqf> |
| Status: |
RESOLVED
FIXED
|
|
|
| Severity: |
normal
|
CC: |
bugwatch, kb9vqf
|
| Priority: |
P5
|
|
|
| Version: |
R14.0.0 [Trinity] | |
|
| Hardware: |
i386 | |
|
| OS: |
Debian Wheezy | |
|
|
Compiler Version:
|
|
TDE Version String:
|
|
|---|
|
Application Version:
|
|
Application Name:
|
|
|---|
A bug in libr0 causes FTBFS when the build occurs in a different filesystem than the filesystem containing /tmp. libr0 is supposed to be able to handle this case but the function to handle this case - safe_rename - has a bug which causes it to always fail. safe_rename is defined in libr-bfd.c. The buggy code in safe_rename is: fclose(in); fclose(out); if(ferror(in)) After fclose(in) any further access to 'in' results in undefined behavior. What I am seeing in tests is that the ferror(in) here is always true. ferror needs to be checked before fclose. There are several other obvious bugs in this function but they don't cause FTBFS - they just make it fragile. For example not checking the result of fread, not checking ferror(out) in the loop and not checking ferror(out) before fclose. Also the middle two arguments of the fwrite call are weird. TTBOMK they work as written but they look odd and should be swapped to match the fread. In a TDE build this causes FTBFS errors when running tdelfeditor such as: opening executable file /cache/wip/T/tdevelop-trinity-14.0.0-r372/obj-i486-linux-gnu/lib/widgets/propeditor/libkdevpropertyeditor.so... failed to rename output file: Invalid cross-device link