On 3/9/22 1:18 AM, Christoph Hellwig wrote:

On Tue, Mar 08, 2022 at 04:38:21PM -0500, Boris Ostrovsky wrote:

quoted

On 3/1/22 5:53 AM, Christoph Hellwig wrote:

quoted

Allow to pass a remap argument to the swiotlb initialization functions
to handle the Xen/x86 remap case.  ARM/ARM64 never did any remapping
from xen_swiotlb_fixup, so we don't even need that quirk.

Any chance this patch could be split? Lots of things are happening here and it's somewhat hard to review. (Patch 7 too BTW but I think I managed to get through it)

What would be your preferred split?

swiotlb_init() rework to be done separately?

quoted

quoted

diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c
index e0def4b1c3181..2f2c468acb955 100644
--- a/arch/x86/kernel/pci-dma.c
+++ b/arch/x86/kernel/pci-dma.c

@@ -71,15 +71,12 @@ static inline void __init pci_swiotlb_detect(void)
   #endif /* CONFIG_SWIOTLB */
     #ifdef CONFIG_SWIOTLB_XEN
-static bool xen_swiotlb;
-
   static void __init pci_xen_swiotlb_init(void)
   {
   	if (!xen_initial_domain() && !x86_swiotlb_enable)
   		return;

Now that there is a single call site for this routine I think this check can be dropped. We are only called here for xen_initial_domain()==true.

The callsite just checks xen_pv_domain() and itself is called
unconditionally during initialization.

Oh, right, nevermind. *pv* domain.


-boris