]> err.no Git - linux-2.6/commit
projects / linux-2.6 / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 5a820fa) | patch
[SPARC64] mm: don't re-evaluate *ptep
authorHugh Dickins <hugh@veritas.com>
Mon, 7 Nov 2005 22:08:46 +0000 (14:08 -0800)
committerDavid S. Miller <davem@davemloft.net>
Mon, 7 Nov 2005 22:08:46 +0000 (14:08 -0800)
commitb8ae48656db860d4c83a29aa7b0588fc89361935
treedbeee0739669c6e77799b5c3a88fa27326647e20tree | snapshot
parent5a820fa7e1a34f12fec4e6766e5c335ae9427028commit | diff
[SPARC64] mm: don't re-evaluate *ptep

sparc64 prom_callback and new_setup_frame32 each operates on a user page
table without holding lock, and no doubt they've good reason.  But I'd
feel more confident if they were to do a "pte = *ptep" and then operate
on pte, rather than re-evaluating *ptep.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
arch/sparc64/kernel/setup.c diff | blob | history
arch/sparc64/kernel/signal32.c diff | blob | history
Linux 2.6 source tree