mmc: core: use non-reentrant workqueue for clock gating

The default multithread workqueue can cause the same work to be executed
concurrently on a different CPUs. This isn't really suitable for clock
gating as it might already gated the clock and gating it twice results both
host->clk_old and host->ios.clock to be set to 0.

To prevent this from happening we use system_nrt_wq instead.

Signed-off-by: Mika Westerberg <[email protected]>
Reviewed-by: Linus Walleij <[email protected]>
Tested-by: Chris Ball <[email protected]>
Cc: <[email protected]>
Signed-off-by: Chris Ball <[email protected]>
This commit is contained in:
Mika Westerberg 2011-08-18 15:23:49 +03:00 • committed by Chris Ball
commit 50a50f9248

View file

@ -179,7 +179,7 @@ void mmc_host_clk_release(struct mmc_host *host)
host->clk_requests--;
if (mmc_host_may_gate_card(host->card) &&
!host->clk_requests)
schedule_work(&host->clk_gate_work);
queue_work(system_nrt_wq, &host->clk_gate_work);
spin_unlock_irqrestore(&host->clk_lock, flags);
}