AcoSail/kernel_optimize_test
8
0
Fork 0
forked from luck/tmp_suning_uos_patched
Code Pull Requests Activity

ARM: 7438/1: fill possible PMD empty section gaps

Browse Source 
On ARM with the 2-level page table format, a PMD entry is represented by
two consecutive section entries covering 2MB of virtual space.

However, static mappings always were allowed to use separate 1MB section
entries.  This means in practice that a static mapping may create half
populated PMDs via create_mapping().

Since commit 0536bdf33f (ARM: move iotable mappings within the vmalloc
region) those static mappings are located in the vmalloc area. We must
ensure no such half populated PMDs are accessible once vmalloc() or
ioremap() start looking at the vmalloc area for nearby free virtual
address ranges, or various things leading to a kernel crash will happen.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
Reported-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Tested-by: "R, Sricharan" <r.sricharan@ti.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This commit is contained in:
Nicolas Pitre 2012-06-27 17:28:57 +01:00 committed by Russell King
parent 9973290ce2
commit 19b52abe3c
1 changed files with 74 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

74
arch/arm/mm/mmu.c
View File

@ -791,6 +791,79 @@ void __init iotable_init(struct map_desc *io_desc, int nr)
}
}
#ifndef CONFIG_ARM_LPAE
/*
* The Linux PMD is made of two consecutive section entries covering 2MB
* (see definition in include/asm/pgtable-2level.h). However a call to
* create_mapping() may optimize static mappings by using individual
* 1MB section mappings. This leaves the actual PMD potentially half
* initialized if the top or bottom section entry isn't used, leaving it
* open to problems if a subsequent ioremap() or vmalloc() tries to use
* the virtual space left free by that unused section entry.
*
* Let's avoid the issue by inserting dummy vm entries covering the unused
* PMD halves once the static mappings are in place.
*/
static void __init pmd_empty_section_gap(unsigned long addr)
{
struct vm_struct *vm;
vm = early_alloc_aligned(sizeof(*vm), __alignof__(*vm));
vm->addr = (void *)addr;
vm->size = SECTION_SIZE;
vm->flags = VM_IOREMAP | VM_ARM_STATIC_MAPPING;
vm->caller = pmd_empty_section_gap;
vm_area_add_early(vm);
}
static void __init fill_pmd_gaps(void)
{
struct vm_struct *vm;
unsigned long addr, next = 0;
pmd_t *pmd;
/* we're still single threaded hence no lock needed here */
for (vm = vmlist; vm; vm = vm->next) {
if (!(vm->flags & VM_ARM_STATIC_MAPPING))
continue;
addr = (unsigned long)vm->addr;
if (addr < next)
continue;
/*
* Check if this vm starts on an odd section boundary.
* If so and the first section entry for this PMD is free
* then we block the corresponding virtual address.
*/
if ((addr & ~PMD_MASK) == SECTION_SIZE) {
pmd = pmd_off_k(addr);
if (pmd_none(*pmd))
pmd_empty_section_gap(addr & PMD_MASK);
}
/*
* Then check if this vm ends on an odd section boundary.
* If so and the second section entry for this PMD is empty
* then we block the corresponding virtual address.
*/
addr += vm->size;
if ((addr & ~PMD_MASK) == SECTION_SIZE) {
pmd = pmd_off_k(addr) + 1;
if (pmd_none(*pmd))
pmd_empty_section_gap(addr);
}
/* no need to look at any vm entry until we hit the next PMD */
next = (addr + PMD_SIZE - 1) & PMD_MASK;
}
}
#else
#define fill_pmd_gaps() do { } while (0)
#endif
static void * __initdata vmalloc_min =
(void *)(VMALLOC_END - (240 << 20) - VMALLOC_OFFSET);
@ -1072,6 +1145,7 @@ static void __init devicemaps_init(struct machine_desc *mdesc)
*/
if (mdesc->map_io)
mdesc->map_io();
fill_pmd_gaps();
/*
* Finally flush the caches and tlb to ensure that we're in a