Detect and solve the overflow on phys_addr_t type for start + size in mmu_set_region_dcache_behaviour() function.
This issue occurs for example with ARM32, start = 0xC0000000 and size = 0x40000000: start + size = 0x100000000 and end = 0x0. Overflow is detected when end < start. In normal case the previous behavior is still used: when start is not aligned on MMU section, the end address is only aligned after the sum start + size. Signed-off-by: Patrick Delaunay <patrick.delau...@st.com> --- arch/arm/lib/cache-cp15.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/arch/arm/lib/cache-cp15.c b/arch/arm/lib/cache-cp15.c index d15144188b..e5a7fd0ef4 100644 --- a/arch/arm/lib/cache-cp15.c +++ b/arch/arm/lib/cache-cp15.c @@ -63,6 +63,11 @@ void mmu_set_region_dcache_behaviour(phys_addr_t start, size_t size, end = ALIGN(start + size, MMU_SECTION_SIZE) >> MMU_SECTION_SHIFT; start = start >> MMU_SECTION_SHIFT; + + /* phys_addr_t overflow detected */ + if (end < start) + end = (~(phys_addr_t)0x0 >> MMU_SECTION_SHIFT) + 1; + #ifdef CONFIG_ARMV7_LPAE debug("%s: start=%pa, size=%zu, option=%llx\n", __func__, &start, size, option); -- 2.17.1