From: catalin.marinas@arm.com (Catalin Marinas) Date: Tue, 28 May 2013 17:04:40 +0100 Subject: [PATCH] ARM: map_init_section flushes incorrect pmd In-Reply-To: <20130528130502.GA28971@mudshark.cambridge.arm.com> References: <1369738100-11152-1-git-send-email-ratbert.chuang@gmail.com> <20130528130502.GA28971@mudshark.cambridge.arm.com> Message-ID: <20130528160440.GF17912@arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Tue, May 28, 2013 at 02:05:02PM +0100, Will Deacon wrote: > On Tue, May 28, 2013 at 11:48:20AM +0100, Po-Yu Chuang wrote: > > This bug was introduced in commit e651eab0. > > Some v4/v5 platforms failed to boot due to this. > > > > Signed-off-by: Po-Yu Chuang > > --- > > arch/arm/mm/mmu.c | 4 +++- > > 1 file changed, 3 insertions(+), 1 deletion(-) > > > > diff --git a/arch/arm/mm/mmu.c b/arch/arm/mm/mmu.c > > index e0d8565..19a43f8 100644 > > --- a/arch/arm/mm/mmu.c > > +++ b/arch/arm/mm/mmu.c > > @@ -620,6 +620,8 @@ static void __init map_init_section(pmd_t *pmd, unsigned long addr, > > unsigned long end, phys_addr_t phys, > > const struct mem_type *type) > > { > > + pmd_t *p = pmd; > > + > > #ifndef CONFIG_ARM_LPAE > > /* > > * In classic MMU format, puds and pmds are folded in to > > @@ -638,7 +640,7 @@ static void __init map_init_section(pmd_t *pmd, unsigned long addr, > > phys += SECTION_SIZE; > > } while (pmd++, addr += SECTION_SIZE, addr != end); > > > > - flush_pmd_entry(pmd); > > + flush_pmd_entry(p); > > Wait, shouldn't this flush be *inside* the loop anyway? Otherwise we just > flush the cacheline containing the first pmd. The flushing code could also > flush to PoU instead of PoC for UP ARMv7, but that's an unrelated optimisation. With LPAE, map_init_section() is called once per section from alloc_init_pmd(). The loop is there for classic MMU to allow setting two pmd entries (maximum) and flush_pmd_entry() takes care of both (it flushes a full cache line anyway). -- Catalin