Re: [PATCH] smp: kill unnecessary variable - csd_flags

2013-07-29 Thread Xie XiuQi
On 2013/7/29 20:13, Peter Zijlstra wrote: > On Mon, Jul 29, 2013 at 10:29:45AM +0800, Xie XiuQi wrote: >> We used csd_flags formerly because we allocated csd_data by >> kmalloc when "wait == 0". When fail to allocation, we will >> fall back to on-stack allocation. "csd_data" might be invalid >> aft

Re: [PATCH] smp: kill unnecessary variable - csd_flags

2013-07-29 Thread Peter Zijlstra
On Mon, Jul 29, 2013 at 10:29:45AM +0800, Xie XiuQi wrote: > We used csd_flags formerly because we allocated csd_data by > kmalloc when "wait == 0". When fail to allocation, we will > fall back to on-stack allocation. "csd_data" might be invalid > after generic_exec_single return. > > But now we u

[PATCH] smp: kill unnecessary variable - csd_flags

2013-07-28 Thread Xie XiuQi
We used csd_flags formerly because we allocated csd_data by kmalloc when "wait == 0". When fail to allocation, we will fall back to on-stack allocation. "csd_data" might be invalid after generic_exec_single return. But now we use per cpu data for single cpu ipi calls, and csd_data can't fall back