Re: [PATCH -next] uprobes: fix two zero old_folio bugs in __replace_page()

[Tong Tiangen]

在 2025/2/18 0:12, Oleg Nesterov 写道:
Can't comment, my understanding of mm/ is not enough these days.

Just one question...

On 02/17, Tong Tiangen wrote:

Fixes: 7396fa818d62 ("uprobes/core: Make background page replacement logic account 
for rss_stat counters")
Fixes: 2b1444983508 ("uprobes, mm, x86: Add the ability to install and remove 
uprobes breakpoints")

Are you sure this logic was wrong from the very beginning? Just curious.

Oleg.

Yes, i checked the original code logic, and the put_page() didn't take 
zero pages into account.

Add Morton,Peter and David for discussion.

Thanks,
Tong.


Signed-off-by: Tong Tiangen <tongtian...@huawei.com>
---
  kernel/events/uprobes.c | 6 ++++--
  1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
index 46ddf3a2334d..ff5694acfa68 100644
--- a/kernel/events/uprobes.c
+++ b/kernel/events/uprobes.c
@@ -213,7 +213,8 @@ static int __replace_page(struct vm_area_struct *vma, 
unsigned long addr,
                dec_mm_counter(mm, MM_ANONPAGES);
  
  	if (!folio_test_anon(old_folio)) {
-               dec_mm_counter(mm, mm_counter_file(old_folio));
+               if (!is_zero_folio(old_folio))
+                       dec_mm_counter(mm, mm_counter_file(old_folio));
                inc_mm_counter(mm, MM_ANONPAGES);
        }
  
@@ -227,7 +228,8 @@ static int __replace_page(struct vm_area_struct *vma, unsigned long addr,
        if (!folio_mapped(old_folio))
                folio_free_swap(old_folio);
        page_vma_mapped_walk_done(&pvmw);
-       folio_put(old_folio);
+       if (!is_zero_folio(old_folio))
+               folio_put(old_folio);
  
  	err = 0;
   unlock:
--
2.25.1



.