VM_IOREMAP is used to access hardware through a mechanism called
I/O mapped memory. Android binder is a IPC machanism which will
not access I/O memory.
And VM_IOREMAP has alignment requiement which may not needed in
binder.
__get_vm_area_node()
{
...
if (flags & VM_IOREMAP)
align = 1ul << clamp_t(int, fls_long(size),
PAGE_SHIFT, IOREMAP_MAX_ORDER);
...
}
This patch will save some kernel vm area, especially for 32bit os.
In 32bit OS, kernel vm area is only 240MB. We may got below
error when launching a app:
<3>[ 4482.440053] binder_alloc: binder_alloc_mmap_handler: 15728 8ce67000-8cf65000 get_vm_area failed -12
<3>[ 4483.218817] binder_alloc: binder_alloc_mmap_handler: 15745 8ce67000-8cf65000 get_vm_area failed -12
Signed-off-by: Ganesh Mahendran <[email protected]>
----
V3: update comments
V2: update comments
---
drivers/android/binder_alloc.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/android/binder_alloc.c b/drivers/android/binder_alloc.c
index 07b866a..5a426c8 100644
--- a/drivers/android/binder_alloc.c
+++ b/drivers/android/binder_alloc.c
@@ -670,7 +670,7 @@ int binder_alloc_mmap_handler(struct binder_alloc *alloc,
goto err_already_mapped;
}
- area = get_vm_area(vma->vm_end - vma->vm_start, VM_IOREMAP);
+ area = get_vm_area(vma->vm_end - vma->vm_start, VM_ALLOC);
if (area == NULL) {
ret = -ENOMEM;
failure_string = "get_vm_area";
--
1.9.1
On Wed, Jan 10, 2018 at 10:49:05AM +0800, Ganesh Mahendran wrote:
> VM_IOREMAP is used to access hardware through a mechanism called
> I/O mapped memory. Android binder is a IPC machanism which will
> not access I/O memory.
>
> And VM_IOREMAP has alignment requiement which may not needed in
> binder.
> __get_vm_area_node()
> {
> ...
> if (flags & VM_IOREMAP)
> align = 1ul << clamp_t(int, fls_long(size),
> PAGE_SHIFT, IOREMAP_MAX_ORDER);
> ...
> }
>
> This patch will save some kernel vm area, especially for 32bit os.
>
> In 32bit OS, kernel vm area is only 240MB. We may got below
> error when launching a app:
>
> <3>[ 4482.440053] binder_alloc: binder_alloc_mmap_handler: 15728 8ce67000-8cf65000 get_vm_area failed -12
> <3>[ 4483.218817] binder_alloc: binder_alloc_mmap_handler: 15745 8ce67000-8cf65000 get_vm_area failed -12
>
> Signed-off-by: Ganesh Mahendran <[email protected]>
> ----
> V3: update comments
> V2: update comments
> ---
> drivers/android/binder_alloc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
Martijn and Todd, any objections to this patch?
thanks,
greg k-h
On Mon, Jan 22, 2018 at 7:54 AM, Greg KH <[email protected]> wrote:
> On Wed, Jan 10, 2018 at 10:49:05AM +0800, Ganesh Mahendran wrote:
>> VM_IOREMAP is used to access hardware through a mechanism called
>> I/O mapped memory. Android binder is a IPC machanism which will
>> not access I/O memory.
>>
>> And VM_IOREMAP has alignment requiement which may not needed in
>> binder.
>> __get_vm_area_node()
>> {
>> ...
>> if (flags & VM_IOREMAP)
>> align = 1ul << clamp_t(int, fls_long(size),
>> PAGE_SHIFT, IOREMAP_MAX_ORDER);
>> ...
>> }
>>
>> This patch will save some kernel vm area, especially for 32bit os.
>>
>> In 32bit OS, kernel vm area is only 240MB. We may got below
>> error when launching a app:
>>
>> <3>[ 4482.440053] binder_alloc: binder_alloc_mmap_handler: 15728 8ce67000-8cf65000 get_vm_area failed -12
>> <3>[ 4483.218817] binder_alloc: binder_alloc_mmap_handler: 15745 8ce67000-8cf65000 get_vm_area failed -12
>>
>> Signed-off-by: Ganesh Mahendran <[email protected]>
>> ----
>> V3: update comments
>> V2: update comments
>> ---
>> drivers/android/binder_alloc.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> Martijn and Todd, any objections to this patch?
Looks fine to me. Arve, do you remember the rationale for using VM_IOREMAP?
>
> thanks,
>
> greg k-h
On Mon, Jan 22, 2018 at 9:02 AM, Todd Kjos <[email protected]> wrote:
> On Mon, Jan 22, 2018 at 7:54 AM, Greg KH <[email protected]> wrote:
>> On Wed, Jan 10, 2018 at 10:49:05AM +0800, Ganesh Mahendran wrote:
>>> VM_IOREMAP is used to access hardware through a mechanism called
>>> I/O mapped memory. Android binder is a IPC machanism which will
>>> not access I/O memory.
>>>
>>> And VM_IOREMAP has alignment requiement which may not needed in
>>> binder.
>>> __get_vm_area_node()
>>> {
>>> ...
>>> if (flags & VM_IOREMAP)
>>> align = 1ul << clamp_t(int, fls_long(size),
>>> PAGE_SHIFT, IOREMAP_MAX_ORDER);
>>> ...
>>> }
>>>
>>> This patch will save some kernel vm area, especially for 32bit os.
>>>
>>> In 32bit OS, kernel vm area is only 240MB. We may got below
>>> error when launching a app:
>>>
>>> <3>[ 4482.440053] binder_alloc: binder_alloc_mmap_handler: 15728 8ce67000-8cf65000 get_vm_area failed -12
>>> <3>[ 4483.218817] binder_alloc: binder_alloc_mmap_handler: 15745 8ce67000-8cf65000 get_vm_area failed -12
>>>
>>> Signed-off-by: Ganesh Mahendran <[email protected]>
>>> ----
>>> V3: update comments
>>> V2: update comments
>>> ---
>>> drivers/android/binder_alloc.c | 2 +-
>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> Martijn and Todd, any objections to this patch?
>
> Looks fine to me. Arve, do you remember the rationale for using VM_IOREMAP?
>
I don't remember for sure, but I think it used alloc_vm_area at some
point, and that uses VM_IOREMAP.
--
Arve Hjønnevåg
Hi, Arve:
2018-01-23 2:55 GMT+08:00 Arve Hjønnevåg <[email protected]>:
> On Mon, Jan 22, 2018 at 9:02 AM, Todd Kjos <[email protected]> wrote:
>> On Mon, Jan 22, 2018 at 7:54 AM, Greg KH <[email protected]> wrote:
>>> On Wed, Jan 10, 2018 at 10:49:05AM +0800, Ganesh Mahendran wrote:
>>>> VM_IOREMAP is used to access hardware through a mechanism called
>>>> I/O mapped memory. Android binder is a IPC machanism which will
>>>> not access I/O memory.
>>>>
>>>> And VM_IOREMAP has alignment requiement which may not needed in
>>>> binder.
>>>> __get_vm_area_node()
>>>> {
>>>> ...
>>>> if (flags & VM_IOREMAP)
>>>> align = 1ul << clamp_t(int, fls_long(size),
>>>> PAGE_SHIFT, IOREMAP_MAX_ORDER);
>>>> ...
>>>> }
>>>>
>>>> This patch will save some kernel vm area, especially for 32bit os.
>>>>
>>>> In 32bit OS, kernel vm area is only 240MB. We may got below
>>>> error when launching a app:
>>>>
>>>> <3>[ 4482.440053] binder_alloc: binder_alloc_mmap_handler: 15728 8ce67000-8cf65000 get_vm_area failed -12
>>>> <3>[ 4483.218817] binder_alloc: binder_alloc_mmap_handler: 15745 8ce67000-8cf65000 get_vm_area failed -12
>>>>
>>>> Signed-off-by: Ganesh Mahendran <[email protected]>
>>>> ----
>>>> V3: update comments
>>>> V2: update comments
>>>> ---
>>>> drivers/android/binder_alloc.c | 2 +-
>>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> Martijn and Todd, any objections to this patch?
>>
>> Looks fine to me. Arve, do you remember the rationale for using VM_IOREMAP?
>>
>
> I don't remember for sure, but I think it used alloc_vm_area at some
> point, and that uses VM_IOREMAP.
Yes, in alloc_vm_area(), it uses VM_IOREMAP flag.
In binder, the ~1MB vm area is cut into binder_buffer which is not
even page aligned.
So for binder IPC, there is no need to align vm area(512KB now).
For 64 bit OS, the vm area is very big, but for 32 bit OS, default vm
area is only 240.
The vm area will soon be exhausted, then no app can be launched.
Thanks.
>
> --
> Arve Hjønnevåg
On Mon, Jan 22, 2018 at 4:54 PM, Greg KH <[email protected]> wrote:
> Martijn and Todd, any objections to this patch?
Looks good to me.
>
> thanks,
>
> greg k-h
Hi, Todd:
2018-01-23 1:02 GMT+08:00 Todd Kjos <[email protected]>:
> On Mon, Jan 22, 2018 at 7:54 AM, Greg KH <[email protected]> wrote:
>> On Wed, Jan 10, 2018 at 10:49:05AM +0800, Ganesh Mahendran wrote:
>>> VM_IOREMAP is used to access hardware through a mechanism called
>>> I/O mapped memory. Android binder is a IPC machanism which will
>>> not access I/O memory.
>>>
>>> And VM_IOREMAP has alignment requiement which may not needed in
>>> binder.
>>> __get_vm_area_node()
>>> {
>>> ...
>>> if (flags & VM_IOREMAP)
>>> align = 1ul << clamp_t(int, fls_long(size),
>>> PAGE_SHIFT, IOREMAP_MAX_ORDER);
>>> ...
>>> }
>>>
>>> This patch will save some kernel vm area, especially for 32bit os.
>>>
>>> In 32bit OS, kernel vm area is only 240MB. We may got below
>>> error when launching a app:
>>>
>>> <3>[ 4482.440053] binder_alloc: binder_alloc_mmap_handler: 15728 8ce67000-8cf65000 get_vm_area failed -12
>>> <3>[ 4483.218817] binder_alloc: binder_alloc_mmap_handler: 15745 8ce67000-8cf65000 get_vm_area failed -12
>>>
>>> Signed-off-by: Ganesh Mahendran <[email protected]>
>>> ----
>>> V3: update comments
>>> V2: update comments
>>> ---
>>> drivers/android/binder_alloc.c | 2 +-
>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> Martijn and Todd, any objections to this patch?
>
> Looks fine to me. Arve, do you remember the rationale for using VM_IOREMAP?
Thanks for your review.
>
>>
>> thanks,
>>
>> greg k-h
Hi, Martijn
2018-01-24 22:33 GMT+08:00 Martijn Coenen <[email protected]>:
> On Mon, Jan 22, 2018 at 4:54 PM, Greg KH <[email protected]> wrote:
>> Martijn and Todd, any objections to this patch?
>
> Looks good to me.
Thanks for your review.
Should I cherry-pick this change to aosp kernel 3.10/3.18/4.4/4.9 now?
Thanks.
>
>>
>> thanks,
>>
>> greg k-h
On Thu, Jan 25, 2018 at 10:09:53AM +0800, Ganesh Mahendran wrote:
> Hi, Martijn
>
> 2018-01-24 22:33 GMT+08:00 Martijn Coenen <[email protected]>:
> > On Mon, Jan 22, 2018 at 4:54 PM, Greg KH <[email protected]> wrote:
> >> Martijn and Todd, any objections to this patch?
> >
> > Looks good to me.
>
> Thanks for your review.
>
> Should I cherry-pick this change to aosp kernel 3.10/3.18/4.4/4.9 now?
If you need it "right now", sure. Otherwise if you can wait 2 weeks, it
will automatically show up in the 3.18/4.4/4.9/4.14 android-common
branches as part of the stable kernel updates.
It's your choice, but either way to get it into 3.10, you will need to
submit it to AOSP.
thanks,
greg k-h