Radix on AIM, and all of Book-E (currently), can do direct addressing of
user space, instead of needing to map user addresses into kernel space.
Take advantage of this to optimize the copy(9) functions for this
behavior, and avoid effectively NOP translations.
Details
Details
- Reviewers
nwhitehorn bdragon - Group Reviewers
PowerPC - Commits
- rS361861: powerpc: Use IFUNCs for copyin/copyout/etc
Tested on powerpcspe, powerpc64/booke, powerpc64/AIM
Diff Detail
Diff Detail
- Repository
- rS FreeBSD src repository - subversion
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
Comment Actions
Fix booke, which doesn't use ifuncs for copyin/copyout. Use a "naked function"
definition instead, since it's really treated as a secondary return path.
Comment Actions
boot tests:
powerpcspe/RB800 ok
powerpc64/X5000 ok
powerpc/G4 ok
powerpc64/POWER9 Radix ok (old elfv1 userland)
powerpc64/POWER9 HPT ok (old elfv1 userland)
Only found comment related issues in readthrough.
| sys/powerpc/powerpc/copyinout.c | ||
|---|---|---|
| 78 ↗ | (On Diff #72752) | minor nit: s/IFUNcs/IFUNCs/ |
| 81 ↗ | (On Diff #72752) | s/powerpc64/AIM powerpc64/ as Book-E is described in the next paragraph |
| 91 ↗ | (On Diff #72752) | The last two paragraphs are overloading the term "direct" -- the rest of the code treat "direct" as in "doesn't map user pointers" instead of "direct" as in "bypasses ifunc" |
| 128 ↗ | (On Diff #72752) | english nit: s/if/whether/ might be easier to read |