Details
- Reviewers
emaste kib jhb - Commits
- rS338274: Configure -zifunc-noplt for amd64 kernels.
Diff Detail
- Repository
- rS FreeBSD src repository - subversion
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 123 ↗ | (On Diff #46785) | aside, I need to add "i386" here before @kib adds i386 ifunc |
| 135 ↗ | (On Diff #46785) | I wonder if we should add a ifunc-noplt to LINKER_FEATURES? ld.bfd will silently ignore the unknown -z option and not use the the optimization, but upstream lld 7 will fail. Hrm, I'm not quite sure how we'd test for our lld 7 vs upstream lld 7 though. I guess we just punt on it. |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 135 ↗ | (On Diff #46785) | Hrmm. Yeah, I don't see a good way to handle this. Hopefully I can get it upstreamed quickly and this will mostly be a non-issue. Do you think it's worth adding a src.conf option or so, so that users can explicitly disable this if they want to use upstream lld? |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 135 ↗ | (On Diff #46785) | I think it's not worth worrying about - let's just try to upstream it as soon as we can. |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 135 ↗ | (On Diff #46785) | although -Wl,z ifunc-noplt seems odd (despite a similar example above), should be -Wl,-z,ifunc-noplt |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 135 ↗ | (On Diff #46785) | I do think LINKER_FEATURES would be the cleanest way to handle this if it isn't hard to detect lld with this change vs lld without. |
| sys/conf/kern.pre.mk | ||
|---|---|---|
| 133 ↗ | (On Diff #46808) | Really, this means that we should be specifying "-z notext". |