Details
- Reviewers
imp andrew - Commits
- rG01028c736cbc: Add a KASAN option to the kernel build
Diff Detail
- Repository
- rS FreeBSD src repository - subversion
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 38125 Build 35014: arc lint + arc unit
Event Timeline
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 110 | I wonder if (as a later change) we should get config(8) to generate a Makefile.config.inc or such that we can include? | |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 110 | I don't quite follow the suggestion. You mean we would end up with a Makefile.KASAN.inc or so? | |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 110 | Something like Makefile.options.inc that contains KASAN_ENABLED=yes and such, rather than invoking grep n times | |
This looks good to me, modulo the refactoring idea it inspired.
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 110 | Hmmm, we've grown a lot of these grep commands. I wrote a generic version of this some time ago (7.x? 8.x?)e. I thought I'd committed it, but it appears that I didn't. Personally, I'd create a mk.$OPTION file in $S/conf. For all options that are enabled that have a mk.$OPTION have config write them into the makefile at %BEFORE_DEPEND. But I'll be the first to admit this is beyond the scope of this patch. | |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 110 | I'll have to add another of these for KMSAN, so a better solution is indeed warranted. The approach of having mk.KASAN seems good. We'd need to audit all of the existing uses of "grep <OPTION> opt_foo.h" to make sure they're all amenable to this approach. | |