On 9/15/21 10:59 PM, Jason Gunthorpe wrote:
On Wed, Sep 15, 2021 at 10:52:40PM +0800, Qi Zheng wrote:I am going to split this patch series as follows: 1. Introduce the new dummy APIs, which is an empty implementation. But I will explain its semantics. 2. Merge #6, #7 and #8, and call these dummy APIs in any necessary location, and split some special cases into single patches, such as pagefault and gup, etc. So that we can explain in more detail the concurrency in these cases. For example, we don't need to hold any pte_refcount in the fast path in gup on the x86_64 platform. Because the PTE page can't be freed after the local CPU interrupt is closed in the fast path in gup. 3. Introduce CONFIG_FREE_USER_PTE and implement these empty dummy APIs. 4. Add a description document. And I try to add a function that combines pte_offset_map() and pte_try_get(). Maybe the func name is pte_try_map() recommended by Jason, or keep the pte_offset_map() unchanged?It is part of the transformation, add a pte_try_map()/pte_undo_try_map() and replace all the pte_offset_map() callsites that can use the new API with it. The idea was that try_map would incorporate the pmd_trans_unstable/etc mess so searching for trans_unstable is a good place to start finding candidates. Some are simple, some are tricky.
Yes, I will search pte_offset_map()/pmd_trans_unstable/etc, and then analyze the specific situation.
When you get to step 3 you just change pte_try_map() and the callsites don't need changing. Jason
Thanks, Qi
