From: Michael Bohan <mbohan@xxxxxxxxxxx>
In FLATMEM configurations, the VM code assumes that memory bank
end addresses in a hole are MAX_ORDER_NR_PAGES aligned if
CONFIG_HOLES_IN_ZONE is not enabled. The crash scenarios are hard
to debug, so let's not proceed if we aren't configured right.
Change-Id: I1494d7bf124e4ec85996af9ae81b394a3322ec6b
Signed-off-by: Michael Bohan <mbohan@xxxxxxxxxxx>
Signed-off-by: Michael Bohan <mbohan@xxxxxxxxxxxxxx>
---
arch/arm/mm/init.c | 18 ++++++++++++++++++
1 files changed, 18 insertions(+), 0 deletions(-)
diff --git a/arch/arm/mm/init.c b/arch/arm/mm/init.c
index ab4890f..3fee7d4 100644
--- a/arch/arm/mm/init.c
+++ b/arch/arm/mm/init.c
@@ -254,6 +254,24 @@ static void __init bootmem_init_node(int node, struct meminfo *mi,
for_each_nodebank(i, mi, node) {
struct membank *bank = &mi->bank[i];
+#if defined(CONFIG_FLATMEM) && !defined(CONFIG_HOLES_IN_ZONE)
+ /*
+ * The VM code assumes that hole end addresses are aligned if
+ * CONFIG_HOLES_IN_ZONE is not enabled. This results in
+ * panics since we free unused memmap entries on ARM.
+ * This check shouldn't be necessary for the last bank's end
+ * address, since the VM code accounts for the total zone size.
+ */
+ if ((i < (mi->nr_banks - 1)) &&
+ (bank_pfn_end(bank) & (MAX_ORDER_NR_PAGES - 1))) {
+ pr_err("Memory bank[%d] not aligned to 0x%x bytes.\n"
+ "\tMake bank end address align with MAX_ORDER\n"
+ "\tor enable option CONFIG_HOLES_IN_ZONE.\n",
+ i, __pfn_to_phys(MAX_ORDER_NR_PAGES));
+ BUG();
+ }
+#endif
+
if (!bank->highmem)
map_memory_bank(bank);
}
--
1.6.2.3
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
