1 Mar
2012
1 Mar
'12
11:30 a.m.
On Thu, Mar 01, 2012 at 05:10:51PM +0800, Ying-Chun Liu (PaulLiu) wrote:
- spin_lock(&adata->reglock);
- val = readl(adata->ioreg + addr);
- spin_unlock(&adata->reglock);
Do you really need to take a lock for a single read operation from a memory mapped register? I'd expect this to be atomic in itself. You need to lock on read/modify/write cycles to make sure that you don't get a read/read/modify/modify/write/write and misplace one of the modifies but that's not an issue for an isolated read.