From ccffff48c93c30733783b05cd1ef843a4c274653 Mon Sep 17 00:00:00 2001 From: Alex Bennée Date: Mon, 4 Apr 2016 15:35:48 +0100 Subject: cpus: don't use atomic_read for vm_clock_warp_start MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit As vm_clock_warp_start is a 64 bit value this causes problems for the compiler trying to come up with a suitable atomic operation on 32 bit hosts. Because the variable is protected by vm_clock_seqlock, we check its value inside a seqlock critical section. Signed-off-by: Alex Bennée Message-Id: <1459780549-12942-2-git-send-email-alex.bennee@linaro.org> Signed-off-by: Paolo Bonzini --- cpus.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/cpus.c b/cpus.c index 8ae477728d..cbeb1f6139 100644 --- a/cpus.c +++ b/cpus.c @@ -338,10 +338,18 @@ static int64_t qemu_icount_round(int64_t count) static void icount_warp_rt(void) { + unsigned seq; + int64_t warp_start; + /* The icount_warp_timer is rescheduled soon after vm_clock_warp_start * changes from -1 to another value, so the race here is okay. */ - if (atomic_read(&vm_clock_warp_start) == -1) { + do { + seq = seqlock_read_begin(&timers_state.vm_clock_seqlock); + warp_start = vm_clock_warp_start; + } while (seqlock_read_retry(&timers_state.vm_clock_seqlock, seq)); + + if (warp_start == -1) { return; } -- cgit v1.2.3