Re: [PATCH 5/8] mm/gup: add FOLL_UNLOCK

[David Hildenbrand <david@redhat.com>]

On 23.01.23 18:58, Jason Gunthorpe wrote:
> On Mon, Jan 23, 2023 at 12:37:17PM +0100, David Hildenbrand wrote:
>> On 17.01.23 16:58, Jason Gunthorpe wrote:
>>> Setting FOLL_UNLOCK allows GUP to lock/unlock the mmap lock on its own. It
>>> is a more explicit replacement for locked != NULL. This clears the way for
>>> passing in locked = 1, without intending that the lock can be unlocked.
>>>
>>> Set the flag in all cases where it is used, eg locked is present in the
>>> external interface or locked is used internally with locked = 0.
>>>
>>> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
>>> ---
>>>    include/linux/mm.h |  1 +
>>>    mm/gup.c           | 31 +++++++++++++++++++------------
>>>    2 files changed, 20 insertions(+), 12 deletions(-)
>>>
>>> diff --git a/include/linux/mm.h b/include/linux/mm.h
>>> index f3f196e4d66d6f..7496a5c8acede1 100644
>>> --- a/include/linux/mm.h
>>> +++ b/include/linux/mm.h
>>> @@ -3089,6 +3089,7 @@ struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
>>>    #define FOLL_FAST_ONLY	0x80000	/* gup_fast: prevent fall-back to slow gup */
>>>    #define FOLL_PCI_P2PDMA	0x100000 /* allow returning PCI P2PDMA pages */
>>>    #define FOLL_INTERRUPTIBLE  0x200000 /* allow interrupts from generic signals */
>>> +#define FOLL_UNLOCK	0x400000 /* allow unlocking the mmap lock */
>>
>> So it's more like "FOLL_UNLOCKABLE" ? FOLL_UNLOCK sounds like an instruction
>> instead.
> 
> Sure, I renamed it
> 
>> Not that I particularly like adding new internal FOLL flags where avoidable
>> :P
> 
> How about, I made a patch to move all the internal FOLL_ flags into
> mm/internal.h ?

That would be a significant improvement. Maybe we could compress each 
set and move the internal ones to the highest bits.

-- 
Thanks,

David / dhildenb