H_PUT_TCE_INDIRECT uses a 4K page with TCEs to allow handling up to 512
TCEs per hypercall. While it is a decent optimization, we rather share
less of secure VM memory so let's avoid sharing.
This only allows H_PUT_TCE_INDIRECT for normal (not secure) VMs.
This keeps using H_STUFF_TCE as it does not require sharing.
Signed-off-by: Alexey Kardashevskiy <aik@xxxxxxxxx>
---
Possible alternatives are:
1. define FW_FEATURE_STUFFTCE (to allow H_STUFF_TCE) in addition to
FW_FEATURE_MULTITCE (make it only enable H_PUT_TCE_INDIRECT) and enable
only FW_FEATURE_STUFFTCE for SVM; pro = no SVM mention in iommu.c,
con = a FW feature bit with very limited use
2. disable FW_FEATURE_MULTITCE and loose H_STUFF_TCE which adds a delay
in booting process
---
arch/powerpc/platforms/pseries/iommu.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/arch/powerpc/platforms/pseries/iommu.c b/arch/powerpc/platforms/pseries/iommu.c
index f6e9b87c82fc..2334a67c7614 100644
--- a/arch/powerpc/platforms/pseries/iommu.c
+++ b/arch/powerpc/platforms/pseries/iommu.c
@@ -192,7 +192,8 @@ static int tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
int ret = 0;
unsigned long flags;
- if ((npages == 1) || !firmware_has_feature(FW_FEATURE_MULTITCE)) {
+ if ((npages == 1) || !firmware_has_feature(FW_FEATURE_MULTITCE) ||
+ is_secure_guest()) {
return tce_build_pSeriesLP(tbl->it_index, tcenum,
tbl->it_page_shift, npages, uaddr,
direction, attrs);
@@ -402,7 +403,8 @@ static int tce_setrange_multi_pSeriesLP(unsigned long start_pfn,
u64 rc = 0;
long l, limit;
- if (!firmware_has_feature(FW_FEATURE_MULTITCE)) {
+ if (!firmware_has_feature(FW_FEATURE_MULTITCE) ||
+ is_secure_guest()) {
unsigned long tceshift = be32_to_cpu(maprange->tce_shift);
unsigned long dmastart = (start_pfn << PAGE_SHIFT) +
be64_to_cpu(maprange->dma_base);
--
2.17.1
