On Fri, Jun 24, 2016 at 04:12:02PM +0200, Ard Biesheuvel wrote:

On 24 June 2016 at 14:01, Dennis Chen [off-list ref] wrote:

quoted

On Fri, Jun 24, 2016 at 12:43:52PM +0200, Ard Biesheuvel wrote:

quoted

On 24 June 2016 at 05:13, Dennis Chen [off-list ref] wrote:

quoted

When booting an ACPI enabled kernel with 'mem=', probably the ACPI data
regions loaded by firmware will beyond the limit of the memory, in this
case we need to nomap the region above the limit while not removing
it from memblock, because once region removed from memblock, the ACPI
will think that region is not a normal memory and map it as device type
memory accordingly. Since the ACPI core will produce non-alignment access
when paring AML data stream, hence result in alignment fault upon the io
mapped memory space.

For example, below is an alignment exception observed on softIron board
when booting the kernel with 'acpi=force mem=8G':
...
[ 0.542475] Unable to handle kernel paging request at virtual address ffff0000080521e7
[ 0.550457] pgd = ffff000008aa0000
[ 0.553880] [ffff0000080521e7] *pgd=000000801fffe003, *pud=000000801fffd003, *pmd=000000801fffc003, *pte=00e80083ff1c1707
[    0.564939] Internal error: Oops: 96000021 [#1] PREEMPT SMP
[    0.570553] Modules linked in:
[    0.573626] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 4.7.0-rc3-next-20160616+ #172
[    0.581344] Hardware name: AMD Overdrive/Supercharger/Default string, BIOS ROD1001A 02/09/2016
[    0.590025] task: ffff800001ef0000 ti: ffff800001ef8000 task.ti: ffff800001ef8000
[    0.597571] PC is at acpi_ns_lookup+0x520/0x734
[    0.602134] LR is at acpi_ns_lookup+0x4a4/0x734
[    0.606693] pc : [<ffff0000083b8b10>] lr : [<ffff0000083b8a94>] pstate: 60000045
[    0.614145] sp : ffff800001efb8b0
[    0.617478] x29: ffff800001efb8c0 x28: 000000000000001b
[    0.622829] x27: 0000000000000001 x26: 0000000000000000
[    0.628181] x25: ffff800001efb9e8 x24: ffff000008a10000
[    0.633531] x23: 0000000000000001 x22: 0000000000000001
[    0.638881] x21: ffff000008724000 x20: 000000000000001b
[    0.644230] x19: ffff0000080521e7 x18: 000000000000000d
[    0.649580] x17: 00000000000038ff x16: 0000000000000002
[    0.654929] x15: 0000000000000007 x14: 0000000000007fff
[    0.660278] x13: ffffff0000000000 x12: 0000000000000018
[    0.665627] x11: 000000001fffd200 x10: 00000000ffffff76
[    0.670978] x9 : 000000000000005f x8 : ffff000008725fa8
[    0.676328] x7 : ffff000008a8df70 x6 : ffff000008a8df70
[    0.681679] x5 : ffff000008a8d000 x4 : 0000000000000010
[    0.687027] x3 : 0000000000000010 x2 : 000000000000000c
[    0.692378] x1 : 0000000000000006 x0 : 0000000000000000
...
[    1.262235] [<ffff0000083b8b10>] acpi_ns_lookup+0x520/0x734
[    1.267845] [<ffff0000083a7160>] acpi_ds_load1_begin_op+0x174/0x4fc
[    1.274156] [<ffff0000083c1f4c>] acpi_ps_build_named_op+0xf8/0x220
[    1.280380] [<ffff0000083c227c>] acpi_ps_create_op+0x208/0x33c
[    1.286254] [<ffff0000083c1820>] acpi_ps_parse_loop+0x204/0x838
[    1.292215] [<ffff0000083c2fd4>] acpi_ps_parse_aml+0x1bc/0x42c
[    1.298090] [<ffff0000083bc6e8>] acpi_ns_one_complete_parse+0x1e8/0x22c
[    1.304753] [<ffff0000083bc7b8>] acpi_ns_parse_table+0x8c/0x128
[    1.310716] [<ffff0000083bb8fc>] acpi_ns_load_table+0xc0/0x1e8
[    1.316591] [<ffff0000083c9068>] acpi_tb_load_namespace+0xf8/0x2e8
[    1.322818] [<ffff000008984128>] acpi_load_tables+0x7c/0x110
[    1.328516] [<ffff000008982ea4>] acpi_init+0x90/0x2c0
[    1.333603] [<ffff0000080819fc>] do_one_initcall+0x38/0x12c
[    1.339215] [<ffff000008960cd4>] kernel_init_freeable+0x148/0x1ec
[    1.345353] [<ffff0000086b7d30>] kernel_init+0x10/0xec
[    1.350529] [<ffff000008084e10>] ret_from_fork+0x10/0x40
[    1.355878] Code: b9009fbc 2a00037b 36380057 3219037b (b9400260)
[    1.362035] ---[ end trace 03381e5eb0a24de4 ]---
[    1.366691] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b

With 'efi=debug', we can see those ACPI regions loaded by firmware on
that board as:
[    0.000000] efi:   0x0083ff1b5000-0x0083ff1c2fff [ACPI Reclaim Memory|   |  |  |  |  |  |  |   |WB|WT|WC|UC]*
[    0.000000] efi:   0x0083ff223000-0x0083ff224fff [ACPI Memory NVS    |   |  |  |  |  |  |  |   |WB|WT|WC|UC]*

This patch is trying to address the above issues by nomaping the region
instead of removing it.

Signed-off-by: Dennis Chen <redacted>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Steve Capper <redacted>
Cc: Ard Biesheuvel <redacted>
Cc: Will Deacon <redacted>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Rafael J. Wysocki <redacted>
Cc: Matt Fleming <redacted>
Cc: linux-mm at kvack.org
Cc: linux-acpi at vger.kernel.org
Cc: linux-efi at vger.kernel.org
---
Changes in v2:
Update the commit message and remove the memblock_is_map_memory() check
according to the suggestion from Mark Rutland.

 arch/arm64/mm/init.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
index d45f862..6af2456 100644
--- a/arch/arm64/mm/init.c
+++ b/arch/arm64/mm/init.c

@@ -222,12 +222,13 @@ void __init arm64_memblock_init(void)

        /*
         * Apply the memory limit if it was set. Since the kernel may be loaded
-        * high up in memory, add back the kernel region that must be accessible
-        * via the linear mapping.
+        * in the memory regions above the limit, so we need to clear the
+        * MEMBLOCK_NOMAP flag of this region to make it can be accessible via
+        * the linear mapping.
         */
        if (memory_limit != (phys_addr_t)ULLONG_MAX) {
-               memblock_enforce_memory_limit(memory_limit);
-               memblock_add(__pa(_text), (u64)(_end - _text));
+               memblock_mem_limit_mark_nomap(memory_limit);
+               memblock_clear_nomap(__pa(_text), (u64)(_end - _text));

Up until now, we have ignored the effect of having NOMAP memblocks on
the return values of functions like memblock_phys_mem_size() and
memblock_mem_size(), since they could reasonably be expected to cover
only a small slice of all available memory. However, after applying
this patch, it may well be the case that most of memory is marked
NOMAP, and these functions will cease to work as expected.

Hi Ard, I noticed these inconsistences as you mentioned, but seems the
available memory is limited correctly. For this case('mem='), will it bring
some substantive side effects except that some log messages maybe confusing?

That is exactly the question that needs answering before we can merge
these patches. I know we consider mem= a development hack, but the
intent is to make it appear to the kernel as if only a smaller amount
of memory is available to the kernel, and this is signficantly
different from having memblock_mem_size() et al return much larger
values than what is actually available. Perhaps this doesn't matter at
all, but it is something we must discuss before proceeding with these
changes.

Indeed. So let's go back to the method below... 

quoted

quoted

This means NOMAP is really only suited to punch some holes into the
kernel direct mapping, and so implementing the memory limit by marking
everything NOMAP is not the way to go. Instead, we should probably
reorder the init sequence so that the regions that are reserved in the
UEFI memory map are declared and marked NOMAP [again] after applying
the memory limit in the old way.

Before this patch, I have another one addressing the same issue [1], with
that patch we'll not have these inconsistences, but it looks like a little
bit complicated, so it becomes current one. Any comments about that?

[1]http://lists.infradead.org/pipermail/linux-arm-kernel/2016-June/438443.html

The problem caused by mem= is that it removes regions that are marked
NOMAP. So instead of marking everything above the limit NOMAP, I would
much rather see an alternative implementation of
memblock_enforce_memory_limit() that enforces the mem= limit by only
removing memblocks that have to NOMAP flag cleared, and leaving the
NOMAP ones where they are.

At least for me, this approach will mitigate the inconsistence in some degree while
keeping the similar logic as it was, so I will post an updated version patch soon.

Thanks,
Dennis