From: Song Liu <songliubraving@fb.com>
To: Oleg Nesterov <oleg@redhat.com>
Cc: lkml <linux-kernel@vger.kernel.org>,
Linux-MM <linux-mm@kvack.org>,
"Andrew Morton" <akpm@linux-foundation.org>,
Matthew Wilcox <matthew.wilcox@oracle.com>,
"Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
Kernel Team <Kernel-team@fb.com>,
"William Kucharski" <william.kucharski@oracle.com>,
"srikar@linux.vnet.ibm.com" <srikar@linux.vnet.ibm.com>
Subject: Re: [PATCH v4 1/2] khugepaged: enable collapse pmd for pte-mapped THP
Date: Tue, 6 Aug 2019 21:03:58 +0000 [thread overview]
Message-ID: <38C0FD76-3CA5-4B86-AB31-FDDD72F6C557@fb.com> (raw)
In-Reply-To: <20190806100256.GA21454@redhat.com>
> On Aug 6, 2019, at 3:02 AM, Oleg Nesterov <oleg@redhat.com> wrote:
>
> On 08/02, Song Liu wrote:
>>
>> +void collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr)
>> +{
>> + unsigned long haddr = addr & HPAGE_PMD_MASK;
>> + struct vm_area_struct *vma = find_vma(mm, haddr);
>> + pmd_t *pmd = mm_find_pmd(mm, haddr);
>> + struct page *hpage = NULL;
>> + spinlock_t *ptl;
>> + int count = 0;
>> + pmd_t _pmd;
>> + int i;
>> +
>> + if (!vma || !vma->vm_file || !pmd ||
>> + vma->vm_start > haddr || vma->vm_end < haddr + HPAGE_PMD_SIZE)
>> + return;
>
> I still can't understand why is it safe to blindly use mm_find_pmd().
>
> Say, what pmd_offset(pud, address) will return to this function if
> pud_huge() == T? IIUC, this is possible if is_file_hugepages(vm_file).
> How the code below can use this result?
IIUC, the concern is matching files in hugetlbfs. Maybe we can exclude
that specifically?
>
> I think you need something like hugepage_vma_check() or even
> hugepage_vma_revalidate().
To use hugepage_vma_check(), we will need to set VM_HUGEPAGE for the
following case:
mount shm with huge=always
copy app to shm
start app and enable uprobe
disable uprobe
This vma will not have VM_HUGEPAGE, so it will fail hugepage_vma_check().
How about something like:
diff --git i/kernel/events/uprobes.c w/kernel/events/uprobes.c
index 94d38a39d72e..f0d3e367f907 100644
--- i/kernel/events/uprobes.c
+++ w/kernel/events/uprobes.c
@@ -532,8 +532,10 @@ int uprobe_write_opcode(struct arch_uprobe *auprobe, struct mm_struct *mm,
put_page(new_page);
new_page = NULL;
- if (PageCompound(orig_page))
+ if (PageCompound(orig_page)) {
orig_page_huge = true;
+ vma->vm_flags |= VM_HUGEPAGE;
+ }
}
put_page(orig_page);
}
Thanks,
Song
next prev parent reply other threads:[~2019-08-06 21:04 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-08-02 23:18 [PATCH v4 0/2] khugepaged: collapse pmd for pte-mapped THP Song Liu
2019-08-02 23:18 ` [PATCH v4 1/2] khugepaged: enable " Song Liu
2019-08-06 10:02 ` Oleg Nesterov
2019-08-06 21:03 ` Song Liu [this message]
2019-08-02 23:18 ` [PATCH v4 2/2] uprobe: collapse THP pmd after removing all uprobes Song Liu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=38C0FD76-3CA5-4B86-AB31-FDDD72F6C557@fb.com \
--to=songliubraving@fb.com \
--cc=Kernel-team@fb.com \
--cc=akpm@linux-foundation.org \
--cc=kirill.shutemov@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=matthew.wilcox@oracle.com \
--cc=oleg@redhat.com \
--cc=srikar@linux.vnet.ibm.com \
--cc=william.kucharski@oracle.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).