LKML Archive on lore.kernel.org
 help / color / Atom feed
From: Gerald Schaefer <gerald.schaefer@de.ibm.com>
To: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: linux-mm@kvack.org, christophe.leroy@c-s.fr,
	Jonathan Corbet <corbet@lwn.net>,
	Andrew Morton <akpm@linux-foundation.org>,
	Mike Rapoport <rppt@linux.ibm.com>,
	Vineet Gupta <vgupta@synopsys.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Will Deacon <will@kernel.org>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	Paul Mackerras <paulus@samba.org>,
	Michael Ellerman <mpe@ellerman.id.au>,
	Heiko Carstens <heiko.carstens@de.ibm.com>,
	Vasily Gorbik <gor@linux.ibm.com>,
	Christian Borntraeger <borntraeger@de.ibm.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
	"H. Peter Anvin" <hpa@zytor.com>,
	"Kirill A . Shutemov" <kirill@shutemov.name>,
	Paul Walmsley <paul.walmsley@sifive.com>,
	Palmer Dabbelt <palmer@dabbelt.com>,
	linux-snps-arc@lists.infradead.org,
	linux-arm-kernel@lists.infradead.org,
	linuxppc-dev@lists.ozlabs.org, linux-s390@vger.kernel.org,
	linux-riscv@lists.infradead.org, x86@kernel.org,
	linux-doc@vger.kernel.org, linux-arch@vger.kernel.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH V2 0/3] mm/debug: Add more arch page table helper tests
Date: Tue, 31 Mar 2020 14:30:59 +0200
Message-ID: <20200331143059.29fca8fa@thinkpad> (raw)
In-Reply-To: <1585027375-9997-1-git-send-email-anshuman.khandual@arm.com>

On Tue, 24 Mar 2020 10:52:52 +0530
Anshuman Khandual <anshuman.khandual@arm.com> wrote:

> This series adds more arch page table helper tests. The new tests here are
> either related to core memory functions and advanced arch pgtable helpers.
> This also creates a documentation file enlisting all expected semantics as
> suggested by Mike Rapoport (https://lkml.org/lkml/2020/1/30/40).
> 
> This series has been tested on arm64 and x86 platforms. There is just one
> expected failure on arm64 that will be fixed when we enable THP migration.
> 
> [   21.741634] WARNING: CPU: 0 PID: 1 at mm/debug_vm_pgtable.c:782
> 
> which corresponds to
> 
> WARN_ON(!pmd_present(pmd_mknotpresent(pmd_mkhuge(pmd))))
> 
> There are many TRANSPARENT_HUGEPAGE and ARCH_HAS_TRANSPARENT_HUGEPAGE_PUD
> ifdefs scattered across the test. But consolidating all the fallback stubs
> is not very straight forward because ARCH_HAS_TRANSPARENT_HUGEPAGE_PUD is
> not explicitly dependent on ARCH_HAS_TRANSPARENT_HUGEPAGE.
> 
> This series has been build tested on many platforms including the ones that
> subscribe the test through ARCH_HAS_DEBUG_VM_PGTABLE.
> 

Hi Anshuman,

thanks for the update. There are a couple of issues on s390, some might
also affect other archs.

1) The pxd_huge_tests are using pxd_set/clear_huge, which defaults to
returning 0 if !CONFIG_HAVE_ARCH_HUGE_VMAP. As result, the checks for
!pxd_test/clear_huge in the pxd_huge_tests will always trigger the
warning. This should affect all archs w/o CONFIG_HAVE_ARCH_HUGE_VMAP.
Could be fixed like this:

@@ -923,8 +923,10 @@ void __init debug_vm_pgtable(void)
        pmd_leaf_tests(pmd_aligned, prot);
        pud_leaf_tests(pud_aligned, prot);
 
-       pmd_huge_tests(pmdp, pmd_aligned, prot);
-       pud_huge_tests(pudp, pud_aligned, prot);
+       if (IS_ENABLED(CONFIG_HAVE_ARCH_HUGE_VMAP)) {
+               pmd_huge_tests(pmdp, pmd_aligned, prot);
+               pud_huge_tests(pudp, pud_aligned, prot);
+       }
 
        pte_savedwrite_tests(pte_aligned, prot);
        pmd_savedwrite_tests(pmd_aligned, prot);

BTW, please add some comments to the various #ifdef/#else stuff, especially
when the different parts are far away and/or nested.

2) The hugetlb_advanced_test will fail because it directly de-references
huge *ptep pointers instead of using huge_ptep_get() for this. We have
very different pagetable entry layout for pte and (large) pmd on s390,
and unfortunately the whole hugetlbfs code is using pte_t instead of pmd_t
like THP. For this reason, huge_ptep_get() was introduced, which will
return a "converted" pte, because directly reading from a *ptep (pointing
to a large pmd) will not return a proper pte. Only ARM has also an
implementation of huge_ptep_get(), so they could be affected, depending
on what exactly they need it for.

Could be fixed like this (the first de-reference is a bit special,
because at that point *ptep does not really point to a large (pmd) entry
yet, it is initially an invalid pte entry, which breaks our huge_ptep_get()
conversion logic. I also added PMD_MASK alignment for RANDOM_ORVALUE,
because we do have some special bits there in our large pmds. It seems
to also work w/o that alignment, but it feels a bit wrong):

@@ -731,26 +731,26 @@ static void __init hugetlb_advanced_test
                                          unsigned long vaddr, pgprot_t prot)
 {
        struct page *page = pfn_to_page(pfn);
-       pte_t pte = READ_ONCE(*ptep);
+       pte_t pte;

-       pte = __pte(pte_val(pte) | RANDOM_ORVALUE);
+       pte = pte_mkhuge(mk_pte_phys(RANDOM_ORVALUE & PMD_MASK, prot));
        set_huge_pte_at(mm, vaddr, ptep, pte);
        barrier();
        WARN_ON(!pte_same(pte, huge_ptep_get(ptep)));
        huge_pte_clear(mm, vaddr, ptep, PMD_SIZE);
-       pte = READ_ONCE(*ptep);
+       pte = huge_ptep_get(ptep);
        WARN_ON(!huge_pte_none(pte));
 
        pte = mk_huge_pte(page, prot);
        set_huge_pte_at(mm, vaddr, ptep, pte);
        huge_ptep_set_wrprotect(mm, vaddr, ptep);
-       pte = READ_ONCE(*ptep);
+       pte = huge_ptep_get(ptep);
        WARN_ON(huge_pte_write(pte));
 
        pte = mk_huge_pte(page, prot);
        set_huge_pte_at(mm, vaddr, ptep, pte);
        huge_ptep_get_and_clear(mm, vaddr, ptep);
-       pte = READ_ONCE(*ptep);
+       pte = huge_ptep_get(ptep);
        WARN_ON(!huge_pte_none(pte));
 
        pte = mk_huge_pte(page, prot);
@@ -759,7 +759,7 @@ static void __init hugetlb_advanced_test
        pte = huge_pte_mkwrite(pte);
        pte = huge_pte_mkdirty(pte);
        huge_ptep_set_access_flags(vma, vaddr, ptep, pte, 1);
-       pte = READ_ONCE(*ptep);
+       pte = huge_ptep_get(ptep);
        WARN_ON(!(huge_pte_write(pte) && huge_pte_dirty(pte)));
 }
 #else

3) The pmd_protnone_tests() has an issue, because it passes a pmd to
pmd_protnone() which has not been marked as large. We check for large
pmd in the s390 implementation of pmd_protnone(), and will fail if a
pmd is not large. We had similar issues before, in other helpers, where
I changed the logic on s390 to not require the pmd large check, but I'm
not so sure in this case. Is there a valid use case for doing
pmd_protnone() on "normal" pmds? Or could this be changed like this:

@@ -537,7 +537,7 @@ static void __init pte_protnone_tests(un
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 static void __init pmd_protnone_tests(unsigned long pfn, pgprot_t prot)
 {
-       pmd_t pmd = pfn_pmd(pfn, prot);
+       pmd_t pmd = mk_huge_pmd(pfn_to_page(pfn), prot);

        if (!IS_ENABLED(CONFIG_NUMA_BALANCING))
                return;

Regards,
Gerald


  parent reply index

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-03-24  5:22 Anshuman Khandual
2020-03-24  5:22 ` [PATCH V2 1/3] mm/debug: Add tests validating arch page table helpers for core features Anshuman Khandual
2020-03-24 13:29   ` Zi Yan
2020-03-26  2:18     ` Anshuman Khandual
2020-03-26 17:08       ` Zi Yan
2020-03-30  8:56   ` [mm/debug] f675f2f91d: WARNING:at_mm/debug_vm_pgtable.c:#debug_vm_pgtable kernel test robot
2020-04-05 14:49     ` Anshuman Khandual
2020-03-24  5:22 ` [PATCH V2 2/3] mm/debug: Add tests validating arch advanced page table helpers Anshuman Khandual
2020-03-27  7:41   ` [mm/debug] d157503f6f: WARNING:at_mm/debug_vm_pgtable.c:#debug_vm_pgtable kernel test robot
2020-03-24  5:22 ` [PATCH V2 3/3] Documentation/mm: Add descriptions for arch page table helpers Anshuman Khandual
2020-03-26  2:23 ` [PATCH V2 0/3] mm/debug: Add more arch page table helper tests Anshuman Khandual
2020-03-26 15:23   ` Christophe Leroy
2020-03-27  6:46     ` Anshuman Khandual
2020-03-27  7:00       ` Christophe Leroy
2020-03-29 14:21         ` Anshuman Khandual
2020-03-31 12:30 ` Gerald Schaefer [this message]
2020-04-05 12:28   ` Anshuman Khandual
2020-04-07 15:54     ` Gerald Schaefer
2020-04-08  7:11       ` Anshuman Khandual
2020-04-08 12:15         ` Gerald Schaefer
2020-04-09  1:06           ` Anshuman Khandual

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=20200331143059.29fca8fa@thinkpad \
    --to=gerald.schaefer@de.ibm.com \
    --cc=akpm@linux-foundation.org \
    --cc=anshuman.khandual@arm.com \
    --cc=benh@kernel.crashing.org \
    --cc=borntraeger@de.ibm.com \
    --cc=bp@alien8.de \
    --cc=catalin.marinas@arm.com \
    --cc=christophe.leroy@c-s.fr \
    --cc=corbet@lwn.net \
    --cc=gor@linux.ibm.com \
    --cc=heiko.carstens@de.ibm.com \
    --cc=hpa@zytor.com \
    --cc=kirill@shutemov.name \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-riscv@lists.infradead.org \
    --cc=linux-s390@vger.kernel.org \
    --cc=linux-snps-arc@lists.infradead.org \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=mingo@redhat.com \
    --cc=mpe@ellerman.id.au \
    --cc=palmer@dabbelt.com \
    --cc=paul.walmsley@sifive.com \
    --cc=paulus@samba.org \
    --cc=rppt@linux.ibm.com \
    --cc=tglx@linutronix.de \
    --cc=vgupta@synopsys.com \
    --cc=will@kernel.org \
    --cc=x86@kernel.org \
    /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

LKML Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lkml.kernel.org/lkml/0 lkml/git/0.git
	git clone --mirror https://lkml.kernel.org/lkml/1 lkml/git/1.git
	git clone --mirror https://lkml.kernel.org/lkml/2 lkml/git/2.git
	git clone --mirror https://lkml.kernel.org/lkml/3 lkml/git/3.git
	git clone --mirror https://lkml.kernel.org/lkml/4 lkml/git/4.git
	git clone --mirror https://lkml.kernel.org/lkml/5 lkml/git/5.git
	git clone --mirror https://lkml.kernel.org/lkml/6 lkml/git/6.git
	git clone --mirror https://lkml.kernel.org/lkml/7 lkml/git/7.git
	git clone --mirror https://lkml.kernel.org/lkml/8 lkml/git/8.git
	git clone --mirror https://lkml.kernel.org/lkml/9 lkml/git/9.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 lkml lkml/ https://lkml.kernel.org/lkml \
		linux-kernel@vger.kernel.org
	public-inbox-index lkml

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-kernel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git