10936 umem_genasm needs to be stricter about segments

Review Request #1837 — Created May 9, 2019 and submitted July 8, 2019, 2:43 p.m.

Information
Owner:	rm
Repository:	illumos-gate
Branch:	master
Bugs:	10936
Depends On:
Commit:	be29d4b...
Reviewers
Groups:	general
People:

Description

10936 umem_genasm needs to be stricter about segments
10937 Clean up umem smatch and cerrwarn
Reviewed by: Cody Peter Mello <melloc@joyent.com>
Reviewed by: Patrick Mooney <patrick.mooney@joyent.com>

Testing Done

Issues

0
0
0
1
All Issues: 1

Description	From	Last Updated
For strict standards compliance, should this not be addr = (uintptr_t)NULL, and the same for the check later? Ship it ...	citrus	June 6, 2019, 2:45 p.m.

Ship it!

```
Ship It!
```

Ship it!

usr/src/lib/libumem/common/vmem.c (Diff revision 1)

Show all issues

For strict standards compliance, should this not be addr = (uintptr_t)NULL, and the same for the check later?

Ship it either way!

rm June 2, 2019, 2:22 p.m.

Can you explain why it needs to be that way? The assignments to addr are either taddr, another uintptr_t or the result of the P2PHASEUP macro which itself is using arguments that have already been cast to a uintptr_t. I'm not sure it makes sense at this point to use the cast NULL, but maybe there's something more intricate going on?

citrus June 6, 2019, 2:46 p.m.

I misunderstood the use of 0 as a sentinel value to indicate when the value had not been updated. Since 0 is never valid in memory, this is fine; sorry for the noise.

You have a pending review.

Review Board 3.0.18

10936 umem_genasm needs to be stricter about segments

Status: Closed (submitted)