On Fri, 2010-11-19 at 19:02 -0800, Stepan Moskovchenko wrote:Nope. If we are here, the pclk will always be non-null, which is something that may not necessarily be said for for AXI clock.+ int ret;You don't need to check if pclk is null ?
+
+ ret = clk_enable(drvdata->pclk);
Will fix in v2.iommu_drvdata = dev_get_drvdata(ctx_drvdata->pdev->dev.parent);Just do,
+ if (!iommu_drvdata)
+ BUG();
BUG_ON(!iommu_drvdata);
The flush_iotlb function needs to handle clock control as well, which means it can now fail. As a result, we now need to check its return value, instead of assuming it succeeded.- __flush_iotlb(domain);What the relationship between this __flush_iotlb() and turning the
+ ret = __flush_iotlb(domain);
clocks on/off.
Will fix in v2.- pr_err("===== WOAH! =====\n");Cleanup right? It doesn't need it's own patch, but you could mention in
the description that you've done "minor cleanups" or something to that
effect.