>From: Menon, Nishanth
>
>> +/* The DPC object, passed to our priority event callback routine: */
>You may want to follow linux coding style. and not loose info from the
>comments.
>
>Documentation/kernel-doc-nano-HOWTO.txt
This file is removed anyway, I can keep that for later comment cleanups though.
>
>
>> +struct DPC_OBJECT {
>> + u32 dwSignature; /* Used for object validation. */
>> + void *pRefData; /* Argument for client's DPC. */
>> + DPC_PROC pfnDPC; /* Client's DPC. */
>> + u32 numRequested; /* Number of requested DPC's. */
>> + u32 numScheduled; /* Number of executed DPC's. */
>> + struct tasklet_struct dpc_tasklet;
>[...]
Removed
>> diff --git a/drivers/dsp/bridge/wmd/io_sm.c b/drivers/dsp/bridge/wmd/io_sm.c
>> index 96a5aa6..60dbc62 100644
>> --- a/drivers/dsp/bridge/wmd/io_sm.c
>> +++ b/drivers/dsp/bridge/wmd/io_sm.c
>> @@ -251,7 +251,26 @@ DSP_STATUS WMD_IO_Create(OUT struct IO_MGR **phIOMgr,
>>
>> if (devType == DSP_UNIT) {
>> /* Create a DPC object: */
>> - status = DPC_Create(&pIOMgr->hDPC, IO_DPC, (void *)pIOMgr);
>> + MEM_AllocObject(pIOMgr->hDPC, struct DPC_OBJECT,
>> + IO_MGRSIGNATURE);
>> + if (pIOMgr->hDPC) {
>> + tasklet_init(&pIOMgr->hDPC->dpc_tasklet,
>> + DPC_DeferredProcedure, (u32)pIOMgr->hDPC);
>> + /* Fill out our DPC Object: */
>> + pIOMgr->hDPC->pRefData = (void *)pIOMgr;
>> + pIOMgr->hDPC->pfnDPC = IO_DPC;
>> + pIOMgr->hDPC->numRequested = 0;
>> + pIOMgr->hDPC->numScheduled = 0;
>> +#ifdef DEBUG
>> + pIOMgr->hDPC->numRequestedMax = 0;
>> + pIOMgr->hDPC->cEntryCount = 0;
>> +#endif
>
>Do you want to do a memset with 0 instead and fill up the deltas?
This was an intermediate step before removing the dpc thing, in the last patch
this looks like
[snip]
+ /* Create an IO DPC */
+ tasklet_init(&pIOMgr->dpc_tasklet, IO_DPC, (u32)pIOMgr);
+
+ /* Initialize DPC counters */
+ pIOMgr->dpc_req = 0;
+ pIOMgr->dpc_sched = 0;
+
+ spin_lock_init(&pIOMgr->dpc_lock);
[snip]
>> +#ifdef DEBUG
>> + if (hIOMgr->hDPC->numRequested >
>> + hIOMgr->hDPC->numScheduled +
>> + hIOMgr->hDPC->numRequestedMax) {
>> + hIOMgr->hDPC->numRequestedMax =
>> + hIOMgr->hDPC->numRequested -
>> + hIOMgr->hDPC->numScheduled;
>a) is this safe?
>b) {} for a one liner ? mebbe the level of indentation is too much?
This was removed.
>
>> + }
>> +#endif
>[...]
>
>> --- a/drivers/dsp/bridge/wmd/ue_deh.c
>> +++ b/drivers/dsp/bridge/wmd/ue_deh.c
>> @@ -91,8 +91,27 @@ DSP_STATUS WMD_DEH_Create(OUT struct DEH_MGR **phDehMgr,
>> status = NTFY_Create(&pDehMgr->hNtfy);
>>
>> /* Create a DPC object. */
>> - status = DPC_Create(&pDehMgr->hMmuFaultDpc, MMU_FaultDpc,
>> - (void *)pDehMgr);
>> + MEM_AllocObject(pDehMgr->hMmuFaultDpc, struct DPC_OBJECT,
>> + SIGNATURE);
>> + if (pDehMgr->hMmuFaultDpc) {
>> + tasklet_init(&pDehMgr->hMmuFaultDpc->dpc_tasklet,
>> + DPC_DeferredProcedure,
>> + (u32)pDehMgr->hMmuFaultDpc);
>> + /* Fill out DPC Object */
>> + pDehMgr->hMmuFaultDpc->pRefData = (void *)pDehMgr;
>> + pDehMgr->hMmuFaultDpc->pfnDPC = MMU_FaultDpc;
>> + pDehMgr->hMmuFaultDpc->numRequested = 0;
>> + pDehMgr->hMmuFaultDpc->numScheduled = 0;
>> +#ifdef DEBUG
>> + pDehMgr->hMmuFaultDpc->numRequestedMax = 0;
>> + pDehMgr->hMmuFaultDpc->cEntryCount = 0;
>> +#endif
>> + spin_lock_init(&pDehMgr->hMmuFaultDpc->dpc_lock);
>> + } else {
>> + DBG_Trace(GT_6CLASS, "DEH DPC Create:
>> DSP_EMEMORY\n");
>> + status = DSP_EMEMORY;
>I think you give up at this point right? why continue the code? wont it
>be easier if your code did:
>if (!pDehMgr->hMmuFaultDpc) {
> DBG_Trace(GT_6CLASS, "DEH DPC Create: DSP_EMEMORY\n");
> return DSP_EMEMORY;
>}
> do rest of code..
>
In the end (patch 9) this looks like:
+ /* Create a MMUfault DPC */
+ tasklet_init(&pDehMgr->dpc_tasklet, MMU_FaultDpc, (u32)pDehMgr);
I tried to split the changes to be cut and paste versions, and then cleaning up
from there... like removing DeferredProcedure and replacing it for the actual
DPC (IO_DPC or MMUfault), then removing DPC structure to avoid x->y->z things,
I was trying to avoid a big patch with all the changes.
Thanks for the comments.
- omar
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
