powerpc/ioda: Set "read" permission when "write" is set

jira LE-1907
Rebuild_History Non-Buildable kernel-3.10.0-514.el7
Rebuild_CHGLOG: - [powerpc] ioda: Set "read" permission when "write" is set (Gustavo Duarte) [1297885]
Rebuild_FUZZ: 92.16%
commit-author Alexey Kardashevskiy <[email protected]>
commit 6ecad91

Quite often drivers set only "write" permission assuming that this
includes "read" permission as well and this works on plenty of
platforms. However IODA2 is strict about this and produces an EEH when
"read" permission is not set and reading happens.

This adds a workaround in the IODA code to always add the "read" bit
when the "write" bit is set.

Fixes: 10b35b2 ("powerpc/powernv: Do not set "read" flag if direction==DMA_NONE")
	Cc: [email protected] # 4.2+
	Cc: Benjamin Herrenschmidt <[email protected]>
	Signed-off-by: Alexey Kardashevskiy <[email protected]>
	Tested-by: Douglas Miller <[email protected]>
	Signed-off-by: Michael Ellerman <[email protected]>
(cherry picked from commit 6ecad91)
	Signed-off-by: Jonathan Maple <[email protected]>

Author: PlaidCat

arch/powerpc/platforms/powernv/pci.c

@@ -610,6 +610,9 @@ int pnv_tce_build(struct iommu_table *tbl, long index, long npages,
 	u64 rpn = __pa(uaddr) >> tbl->it_page_shift;
 	long i;
 
+	if (proto_tce & TCE_PCI_WRITE)
+		proto_tce |= TCE_PCI_READ;
+
 	for (i = 0; i < npages; i++) {
 		unsigned long newtce = proto_tce |
 			((rpn + i) << tbl->it_page_shift);
@@ -631,6 +634,9 @@ int pnv_tce_xchg(struct iommu_table *tbl, long index,
 
 	BUG_ON(*hpa & ~IOMMU_PAGE_MASK(tbl));
 
+	if (newtce & TCE_PCI_WRITE)
+		newtce |= TCE_PCI_READ;
+
 	oldtce = xchg(pnv_tce(tbl, idx), cpu_to_be64(newtce));
 	*hpa = be64_to_cpu(oldtce) & ~(TCE_PCI_READ | TCE_PCI_WRITE);
 	*direction = iommu_tce_direction(oldtce);