Fix #33, Resolve print format mismatches on 32-bit host #36
Conversation
|
@astrogeco ready for CCB (no label yet) |
Thanks for the tag, I added the label |
| if (Verbose) printf(" sh_size = 0x%08lx\n", SectionHeader->Shdr64.sh_size); | ||
| if (Verbose) printf(" sh_addr = 0x%lx\n", (long unsigned int)SectionHeader->Shdr64.sh_addr); | ||
| if (Verbose) printf(" sh_offset = 0x%08lx\n", (long unsigned int)SectionHeader->Shdr64.sh_offset); | ||
| if (Verbose) printf(" sh_size = 0x%08lx\n", (long unsigned int)SectionHeader->Shdr64.sh_size); | ||
| if (Verbose) printf(" sh_link = 0x%08x\n", SectionHeader->Shdr64.sh_link); | ||
| if (Verbose) printf(" sh_info = 0x%08x\n", SectionHeader->Shdr64.sh_info); |
There was a problem hiding this comment.
For instance wouldn't the "sh_link" and "sh_info" need the same cast? (these are also words of different lengths depending on arch/binary format).
There was a problem hiding this comment.
Clarified pull request, urgent fix is for vxworks6.9, MCP750. As you mention, likely requires more casts for other systems but not a certification priority.
There was a problem hiding this comment.
How is this related to VxWorks exactly? The tool runs on the host machine, not on VxWorks. Are the mismatches specific to running it on a 32-bit host machine vs 64-bit host machine?
There was a problem hiding this comment.
True enough, this is a 32 bit host. I only fixed the issues that showed up for the vxworks build on the 32 bit host. I should have better described the test system.
There was a problem hiding this comment.
OK, that makes more sense. But I'm still not an advocate of just minimally squelching the visible warnings on a specific 32-bit host (IIRC the machine with the VxWorks 6.9 tools is a 32-bit RHEL 6.x host). Other compilers/libraries might flag other printfs. Unless we actually add all the needed casts, warnings are likely reappear again as soon as someone builds on another system.
There was a problem hiding this comment.
It's a step in the right direction, and removes a cert blocker. Making it work across compilers/libraries is a very valid open source concern, but not my priority.
skliper
changed the title
|
With the the pull request title change this is even more unclear -- this tool doesn't have anything to do with VxWorks .... this tool is compiled for and runs on the build host no matter what the target OS is. |
|
Good point. Likely an indication of too little time and too much work. |
skliper
changed the title
|
20200422 CCB - APPROVED |
Describe the contribution
Casts where needed for print formatting on 32-bit host
Fix #33
Testing performed
Steps taken to test the contribution:
Expected behavior changes
Builds for vxworks w/ 32-bit host
System(s) tested on
Additional context
None (this isn't the only issue on vxworks, but haven't written them up yet)
Third party code
None
Contributor Info - All information REQUIRED for consideration of pull request
Jacob Hageman - NASA/GSFC