Closed waveform80 closed 4 years ago
Hi @waveform80 This overlaps with #631 and #635 .
I'm not convinced by the gcc change, but that is what it is. The original behaviour of printing "string is (null)" is far more useful than a bundle of build warnings.
I want to confirm, but I'm fairly certain that vc_filesys is totally unused (I think it's part of the VPU accessing the host filing system), so we're better off just killing it.
The vc_hostfs code (which is Linux user-space support for file accesses from the VPU) hasn't been used on a Pi in my time here, and possible never.
The vc_hostfs code (which is Linux user-space support for file accesses from the VPU) hasn't been used on a Pi in my time here, and possible never.
Thanks for the confirmation - that was my understanding too. Time for a nuke then.
The vc_hostfs code (which is Linux user-space support for file accesses from the VPU) hasn't been used on a Pi in my time here, and possible never.
Thanks for the confirmation - that was my understanding too. Time for a nuke then.
Even better :) Less code is always nice!
While building userland for the Ubuntu packaging under gcc-10, I encountered the following error:
This patch simply moves the DEBUG_MINOR call into the
if(path)
block to avoid this possibility, though it does mean that in the case of an out-of-memory condition (wherestrdup
has returned NULL) that nothing is output by DEBUG_MINOR. I'm happy to re-jig this to call DEBUG_MINOR in both cases (e.g. with an alternate output message when path is NULL) if required.