diff options
| author | Philip Guenther <guenther@cvs.openbsd.org> | 2023-09-05 05:08:27 +0000 |
|---|---|---|
| committer | Philip Guenther <guenther@cvs.openbsd.org> | 2023-09-05 05:08:27 +0000 |
| commit | e3de810958ed798f8cc17a70a5c8063cd313e250 (patch) | |
| tree | 0cc5b10d9e86da04966d3df5f4aa4b53ae2a8dde /sys | |
| parent | d5e085ac22742797fafc235cbe068ba005a3d938 (diff) | |
Address the case 2b version of inconsistent view across threads of
a page undergoing copy-on-write faulting. We fixed the case 1b
version in rev 1.125 (2022-02-01), but missed this other path.
jsg@ noted that in NetBSD Chuck Silvers had a relevant commit, their
rev 1.234 (2023-08-13), which looks like it fixed both cases due
to their refactoring of common code into a uvmfault_promote()
function.
ok mpi@ jca@
Diffstat (limited to 'sys')
| -rw-r--r-- | sys/uvm/uvm_fault.c | 17 |
1 files changed, 15 insertions, 2 deletions
diff --git a/sys/uvm/uvm_fault.c b/sys/uvm/uvm_fault.c index 28fa1da147c..1ae20b5af75 100644 --- a/sys/uvm/uvm_fault.c +++ b/sys/uvm/uvm_fault.c @@ -1,4 +1,4 @@ -/* $OpenBSD: uvm_fault.c,v 1.134 2023/09/02 08:24:40 mpi Exp $ */ +/* $OpenBSD: uvm_fault.c,v 1.135 2023/09/05 05:08:26 guenther Exp $ */ /* $NetBSD: uvm_fault.c,v 1.51 2000/08/06 00:22:53 thorpej Exp $ */ /* @@ -1455,7 +1455,20 @@ uvm_fault_lower(struct uvm_faultinfo *ufi, struct uvm_faultctx *flt, */ if ((amap_flags(amap) & AMAP_SHARED) != 0) { pmap_page_protect(uobjpage, PROT_NONE); - } + } +#if defined(MULTIPROCESSOR) && !defined(__HAVE_PMAP_MPSAFE_ENTER_COW) + /* + * Otherwise: + * If there are multiple threads, either uvm or the + * pmap has to make sure no threads see the old RO + * mapping once any have seen the new RW mapping. + * uvm does it here by forcing it to PROT_NONE before + * inserting the new mapping. + */ + else if (P_HASSIBLING(curproc)) { + pmap_page_protect(uobjpage, PROT_NONE); + } +#endif /* dispose of uobjpage. drop handle to uobj as well. */ if (uobjpage->pg_flags & PG_WANTED) |