From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by smtp.lore.kernel.org (Postfix) with ESMTP id 9FF31CD11C2 for ; Fri, 5 Apr 2024 12:58:22 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 206186B0178; Fri, 5 Apr 2024 08:58:22 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 1B66A6B0179; Fri, 5 Apr 2024 08:58:22 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 057F16B017A; Fri, 5 Apr 2024 08:58:22 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0017.hostedemail.com [216.40.44.17]) by kanga.kvack.org (Postfix) with ESMTP id D61C16B0178 for ; Fri, 5 Apr 2024 08:58:21 -0400 (EDT) Received: from smtpin08.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay09.hostedemail.com (Postfix) with ESMTP id 8F98E80722 for ; Fri, 5 Apr 2024 12:58:21 +0000 (UTC) X-FDA: 81975481602.08.CF46C9E Received: from mail-lf1-f54.google.com (mail-lf1-f54.google.com [209.85.167.54]) by imf02.hostedemail.com (Postfix) with ESMTP id B0BEE8000E for ; Fri, 5 Apr 2024 12:58:19 +0000 (UTC) Authentication-Results: imf02.hostedemail.com; dkim=pass header.d=gmail.com header.s=20230601 header.b=Zfyu9kV4; dmarc=pass (policy=none) header.from=gmail.com; spf=pass (imf02.hostedemail.com: domain of ioworker0@gmail.com designates 209.85.167.54 as permitted sender) smtp.mailfrom=ioworker0@gmail.com ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1712321899; h=from:from:sender:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references:dkim-signature; bh=uvG8ftiblG7YqudDTfPBSWyIzjLPS3+rvJ5EDGpa490=; b=b8EmDoTd2Co0mOs7GMZ4r0u7v5X35l12N3kcpXnrJAFYk9bQ335oAcAoGHheHOk1rw6ROi Psb7gqGbr3TZjA8Ed4oS855qQqmP8bzboyoRqciGU0jD6F+Gi4PvLnf2Ru0MuKgO9Vm/vJ 9Q4/t1zRbeKebMe8DQlOFUnB0gWoXFM= ARC-Authentication-Results: i=1; imf02.hostedemail.com; dkim=pass header.d=gmail.com header.s=20230601 header.b=Zfyu9kV4; dmarc=pass (policy=none) header.from=gmail.com; spf=pass (imf02.hostedemail.com: domain of ioworker0@gmail.com designates 209.85.167.54 as permitted sender) smtp.mailfrom=ioworker0@gmail.com ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1712321899; a=rsa-sha256; cv=none; b=ATlDVCumKUZrBtd9wmXpNkabcnf2sLYJdFFtkGN6WAWJ/fWtsN+9q9Z0rWMTYniYKt7c7t F1khIR3gDeqgOFTXxAwKRfZaNFofihUGMSt00fZT1LzNiYkQhI1Cm2CYAcym4ZcquM0OuG dnScAbfbhwgxzY1EUwIML4ziiIBm3nM= Received: by mail-lf1-f54.google.com with SMTP id 2adb3069b0e04-516d3776334so1041855e87.1 for ; Fri, 05 Apr 2024 05:58:19 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1712321898; x=1712926698; darn=kvack.org; h=content-transfer-encoding:cc:to:subject:message-id:date:from :in-reply-to:references:mime-version:from:to:cc:subject:date :message-id:reply-to; bh=uvG8ftiblG7YqudDTfPBSWyIzjLPS3+rvJ5EDGpa490=; b=Zfyu9kV4+iexgQ1CXEAYBXEGeJt9r+2lKO2svThLbZinaehkC5s1qgp9Z3WPBr1hdB kf+hcXp4bjn5Hv9Ml3vG4Xgc9LyeyNyAEe9aFvHaD+4djyHLVohOXjGJWtzUBK5dJyiL uKdrFrMtiiPni+OZOTDPd3g0iNTQtQ2U4zAO3yVGImYROe4sqj3waEzJsh9ErSSA5Mx3 f04ig711v8DHK1Es6jRNoRPe+Qb+zgu0YKnDLf6MYXmeDp77wTbDA2734Irg31yD3TKD m7Dt9u9SemCGZft5xdwwvqqFXumynLijhYU0GOBupOnGkONwFj6JisuzVgukF/4fSMXn hXDQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1712321898; x=1712926698; h=content-transfer-encoding:cc:to:subject:message-id:date:from :in-reply-to:references:mime-version:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=uvG8ftiblG7YqudDTfPBSWyIzjLPS3+rvJ5EDGpa490=; b=jtG2j0uHJH+UC0xM/BoYw1B3ZCY9GerxST84tXm3VvsJCeTf/vAYaQXmO+P/lCBJt9 G2Xicu9IriprDG3niz6lTZKJiLBzFJo9vWXybhGkCEvwQiqmi7FqgY0Rv0Fg3BlmhZW1 2hC71SO5GiZtiOCVzPLGVjJw06nnQ7LAEAp6IUlh8ClmNdBEEjw62GpJlWV1JJtO8DBK Cq8oBkfqXJz/Ea/KEK6gRL8jDKkP8YG3trcs9a5VYBZjwG/ARniRxo44rk/YgE1UJCpp NgHiCL51OgDbGaCj5LficjscapTzvSXMqqzQcPLz6+TnzpPGu2x7hJIZVJf5n70b/BFG yKSQ== X-Forwarded-Encrypted: i=1; AJvYcCVhGUz3D+2D6/ST8RNSIKL21/WPbE2SAJxREzcVnJRqOtnZTcwi3m+iFgYsozgHV9lDXSTqacDmeeMaH+uQG6KiX50= X-Gm-Message-State: AOJu0YwtKueKVh/7imclg+1659vRVHUaE/q7RdyNV/5sIjcq1Iy13O0P vcyH24ItmQvYyWoni3ZwEoblRKt+PX3rVsUFkDNlfcIK9So9zbuUClu046SfsC12L8MjTB1ceuA OKDY0JoMQYSD4JvPWfxHRooNLDOI= X-Google-Smtp-Source: AGHT+IGwkGQiDhcs8HfYt/EMSmyeQZ1iWrzZlzK7RNy+fRBAKTtk/TMr4YMlk/B16yMIcMnvmBUVsz4EDZz/qeNrd7Q= X-Received: by 2002:ac2:592c:0:b0:515:c8da:c96d with SMTP id v12-20020ac2592c000000b00515c8dac96dmr1107487lfi.19.1712321897384; Fri, 05 Apr 2024 05:58:17 -0700 (PDT) MIME-Version: 1.0 References: <20240402124029.47846-1-ioworker0@gmail.com> <20240402124029.47846-3-ioworker0@gmail.com> <99c745a1-fe87-4f63-9475-a0a598c92c8c@arm.com> In-Reply-To: <99c745a1-fe87-4f63-9475-a0a598c92c8c@arm.com> From: Lance Yang Date: Fri, 5 Apr 2024 20:58:05 +0800 Message-ID: Subject: Re: [PATCH v4 2/2] mm/madvise: optimize lazyfreeing with mTHP in madvise_free To: Ryan Roberts Cc: akpm@linux-foundation.org, zokeefe@google.com, 21cnbao@gmail.com, shy828301@gmail.com, david@redhat.com, mhocko@suse.com, fengwei.yin@intel.com, xiehuan09@gmail.com, wangkefeng.wang@huawei.com, songmuchun@bytedance.com, peterx@redhat.com, minchan@kernel.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable X-Rspam-User: X-Stat-Signature: o15eppatcgd4om4c49eeuj9wnzo7ruqe X-Rspamd-Server: rspam07 X-Rspamd-Queue-Id: B0BEE8000E X-HE-Tag: 1712321899-701251 X-HE-Meta: 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 t7s95cpT uKB91qU3C4vQqLnt+l0G/a9Yyyw6RJ+nVSxgSgMk0lHrH91LKxitD1OZo2l6MUUQPBpF0nRHeXl/e0x8QGOHFTja8IQ== X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: List-Subscribe: List-Unsubscribe: Hey Ryan, Thanks for taking time to review! On Thu, Apr 4, 2024 at 1:35=E2=80=AFAM Ryan Roberts = wrote: > > On 02/04/2024 13:40, Lance Yang wrote: > > This patch optimizes lazyfreeing with PTE-mapped mTHP[1] > > (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio > > splitting if the large folio is fully mapped within the target range. > > > > If a large folio is locked or shared, or if we fail to split it, we jus= t > > leave it in place and advance to the next PTE in the range. But note th= at > > the behavior is changed; previously, any failure of this sort would cau= se > > the entire operation to give up. As large folios become more common, > > sticking to the old way could result in wasted opportunities. > > > > On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios = of > > the same size results in the following runtimes for madvise(MADV_FREE) = in > > seconds (shorter is better): > > > > Folio Size | Old | New | Change > > ------------------------------------------ > > 4KiB | 0.590251 | 0.590259 | 0% > > 16KiB | 2.990447 | 0.185655 | -94% > > 32KiB | 2.547831 | 0.104870 | -95% > > 64KiB | 2.457796 | 0.052812 | -97% > > 128KiB | 2.281034 | 0.032777 | -99% > > 256KiB | 2.230387 | 0.017496 | -99% > > 512KiB | 2.189106 | 0.010781 | -99% > > 1024KiB | 2.183949 | 0.007753 | -99% > > 2048KiB | 0.002799 | 0.002804 | 0% > > I'm guessing the reason that 2M is not showing any change is because its > PMD-mapped and splitting is already elided? If you were to force it to be Your guess is correct. The lack of change in 2M is because it's PMD-mapped. > PTE-mapped then you'll see the very impressive speed-up there too. Don't = worry > about doing that on my account though; these results are already sufficie= nt IMHO. > > > > > [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm= .com > > [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redh= at.com > > > > Signed-off-by: Lance Yang > > --- > > mm/internal.h | 12 ++++- > > mm/madvise.c | 147 ++++++++++++++++++++++++++------------------------ > > mm/memory.c | 4 +- > > 3 files changed, 88 insertions(+), 75 deletions(-) > > > > diff --git a/mm/internal.h b/mm/internal.h > > index 3df06a152ff0..cdc6e2162b30 100644 > > --- a/mm/internal.h > > +++ b/mm/internal.h > > @@ -132,6 +132,8 @@ static inline pte_t __pte_batch_clear_ignored(pte_t= pte, fpb_t flags) > > * first one is writable. > > * @any_young: Optional pointer to indicate whether any entry except t= he > > * first one is young. > > + * @any_dirty: Optional pointer to indicate whether any entry except t= he > > + * first one is dirty. > > * > > * Detect a PTE batch: consecutive (present) PTEs that map consecutive > > * pages of the same large folio. > > @@ -147,18 +149,20 @@ static inline pte_t __pte_batch_clear_ignored(pte= _t pte, fpb_t flags) > > */ > > static inline int folio_pte_batch(struct folio *folio, unsigned long a= ddr, > > pte_t *start_ptep, pte_t pte, int max_nr, fpb_t flags, > > - bool *any_writable, bool *any_young) > > + bool *any_writable, bool *any_young, bool *any_dirty) > > { > > unsigned long folio_end_pfn =3D folio_pfn(folio) + folio_nr_pages= (folio); > > const pte_t *end_ptep =3D start_ptep + max_nr; > > pte_t expected_pte, *ptep; > > - bool writable, young; > > + bool writable, young, dirty; > > int nr; > > > > if (any_writable) > > *any_writable =3D false; > > if (any_young) > > *any_young =3D false; > > + if (any_dirty) > > + *any_dirty =3D false; > > > > VM_WARN_ON_FOLIO(!pte_present(pte), folio); > > VM_WARN_ON_FOLIO(!folio_test_large(folio) || max_nr < 1, folio); > > @@ -174,6 +178,8 @@ static inline int folio_pte_batch(struct folio *fol= io, unsigned long addr, > > writable =3D !!pte_write(pte); > > if (any_young) > > young =3D !!pte_young(pte); > > + if (any_dirty) > > + dirty =3D !!pte_dirty(pte); > > pte =3D __pte_batch_clear_ignored(pte, flags); > > > > if (!pte_same(pte, expected_pte)) > > @@ -191,6 +197,8 @@ static inline int folio_pte_batch(struct folio *fol= io, unsigned long addr, > > *any_writable |=3D writable; > > if (any_young) > > *any_young |=3D young; > > + if (any_dirty) > > + *any_dirty |=3D dirty; > > > > nr =3D pte_batch_hint(ptep, pte); > > expected_pte =3D pte_advance_pfn(expected_pte, nr); > > diff --git a/mm/madvise.c b/mm/madvise.c > > index bd00b83e7c50..8197effd9f14 100644 > > --- a/mm/madvise.c > > +++ b/mm/madvise.c > > @@ -321,6 +321,38 @@ static inline bool can_do_file_pageout(struct vm_a= rea_struct *vma) > > file_permission(vma->vm_file, MAY_WRITE) =3D=3D 0; > > } > > > > +static inline int madvise_folio_pte_batch(unsigned long addr, unsigned= long end, > > + struct folio *folio, pte_t *pte= , > > + bool *any_writable, bool *any_y= oung, bool *any_dirty) > > any_writable is always NULL. Do you need it? Thanks for pointing that out. It seems that the any_writable parameter is redundant here, so I'll drop it= . > > > +{ > > + int max_nr =3D (end - addr) / PAGE_SIZE; > > + const fpb_t fpb_flags =3D FPB_IGNORE_DIRTY | FPB_IGNORE_SOFT_DIRT= Y; > > + > > + return folio_pte_batch(folio, addr, pte, ptep_get(pte), max_nr, > > ptep_get() was problematic for performance of the order-0 folio case when= we > batched fork(). So we are deliberately passing around the value we alread= y read > in the main loop. Granted this case is not so performance critical becaus= e we > only end up here for large folios. But I would still prefer to just pass = the > data we have already read into this function rather than reading it again= . Thanks for the explanation! I completely agree. I=E2=80=98ll pass the data we=E2=80=99ve already read into this function. > > > + fpb_flags, any_writable, any_young, any_di= rty); > > +} > > + > > +static inline bool madvise_pte_split_folio(struct mm_struct *mm, pmd_t= *pmd, > > + unsigned long addr, struct fol= io *folio, pte_t **pte, > > nit: I know 80 chars is a soft limit now (and I think 100 is a hard limit= ), but > try to be consistent. You could move the addr param to the previous line = and be > within the 100 char limit. Personally I would just make the prototype fit= in 80 > chars (same goes for madvise_folio_pte_batch). Got it. Thanks. > > > + spinlock_t **ptl) > > +{ > > + int err; > > + > > + if (!folio_trylock(folio)) > > + return false; > > + > > + folio_get(folio); > > + pte_unmap_unlock(*pte, *ptl); > > + *pte =3D NULL; > > nit: you don't need this since you are later unconditionally setting it a= gain. Nice. I'll remove it. > > > + err =3D split_folio(folio); > > + folio_unlock(folio); > > + folio_put(folio); > > + > > + *pte =3D pte_offset_map_lock(mm, pmd, addr, ptl); > > + > > + return err =3D=3D 0; > > +} > > + > > static int madvise_cold_or_pageout_pte_range(pmd_t *pmd, > > unsigned long addr, unsigned long end, > > struct mm_walk *walk) > > @@ -456,40 +488,26 @@ static int madvise_cold_or_pageout_pte_range(pmd_= t *pmd, > > * next pte in the range. > > */ > > if (folio_test_large(folio)) { > > - const fpb_t fpb_flags =3D FPB_IGNORE_DIRTY | > > - FPB_IGNORE_SOFT_DIRTY; > > - int max_nr =3D (end - addr) / PAGE_SIZE; > > bool any_young; > > - > > - nr =3D folio_pte_batch(folio, addr, pte, ptent, m= ax_nr, > > - fpb_flags, NULL, &any_young)= ; > > + nr =3D madvise_folio_pte_batch(addr, end, folio, = pte, > > + NULL, &any_young, NU= LL); > > if (any_young) > > ptent =3D pte_mkyoung(ptent); > > > > if (nr < folio_nr_pages(folio)) { > > - int err; > > - > > if (folio_likely_mapped_shared(folio)) > > continue; > > if (pageout_anon_only_filter && !folio_te= st_anon(folio)) > > continue; > > - if (!folio_trylock(folio)) > > - continue; > > - folio_get(folio); > > + > > arch_leave_lazy_mmu_mode(); > > - pte_unmap_unlock(start_pte, ptl); > > - start_pte =3D NULL; > > - err =3D split_folio(folio); > > - folio_unlock(folio); > > - folio_put(folio); > > - if (err) > > - continue; > > - start_pte =3D pte =3D > > - pte_offset_map_lock(mm, pmd, addr= , &ptl); > > + if (madvise_pte_split_folio(mm, pmd, addr= , > > + folio, &start= _pte, &ptl)) > > + nr =3D 0; > > if (!start_pte) > > break; > > + pte =3D start_pte; > > arch_enter_lazy_mmu_mode(); > > - nr =3D 0; > > continue; > > This change fixes a bug I've introduced in my swap-out series. Nice. I tr= ied to > fix in v6, but looking at this, I've realised its still broken. I've repl= ied > against that series with the fix. > > > > } > > } > > @@ -688,72 +706,59 @@ static int madvise_free_pte_range(pmd_t *pmd, uns= igned long addr, > > continue; > > > > /* > > - * If pmd isn't transhuge but the folio is large and > > - * is owned by only this process, split it and > > - * deactivate all pages. > > + * If we encounter a large folio, only split it if it is = not > > + * fully mapped within the range we are operating on. Oth= erwise > > + * leave it as is so that it can be marked as lazyfree. I= f we > > + * fail to split a folio, leave it in place and advance t= o the > > + * next pte in the range. > > */ > > if (folio_test_large(folio)) { > > - int err; > > + bool any_young, any_dirty; > > + nr =3D madvise_folio_pte_batch(addr, end, folio, = pte, > > + NULL, &any_young, &a= ny_dirty); > > + if (any_young || any_dirty) > > + ptent =3D pte_mkdirty(pte_mkyoung(ptent))= ; > > I don't think it makes any difference to how ptent is consumed below, but= its > probably more intuitive to separate these two operations: > > if (any_young) > ptent =3D pte_mkyoung(ptent); > if (any_dirty) > ptent =3D pte_mkdirty(ptent); I agree that it's more intuitive to separate these two operations. Thanks. > > > > > - if (folio_likely_mapped_shared(folio)) > > - break; > > - if (!folio_trylock(folio)) > > - break; > > - folio_get(folio); > > - arch_leave_lazy_mmu_mode(); > > - pte_unmap_unlock(start_pte, ptl); > > - start_pte =3D NULL; > > - err =3D split_folio(folio); > > - folio_unlock(folio); > > - folio_put(folio); > > - if (err) > > - break; > > - start_pte =3D pte =3D > > - pte_offset_map_lock(mm, pmd, addr, &ptl); > > - if (!start_pte) > > - break; > > - arch_enter_lazy_mmu_mode(); > > - pte--; > > - addr -=3D PAGE_SIZE; > > - continue; > > - } > > + if (nr < folio_nr_pages(folio)) { > > + if (folio_likely_mapped_shared(folio)) > > + continue; > > > > - if (folio_test_swapcache(folio) || folio_test_dirty(folio= )) { > > - if (!folio_trylock(folio)) > > - continue; > > - /* > > - * If folio is shared with others, we mustn't cle= ar > > - * the folio's dirty flag. > > - */ > > - if (folio_mapcount(folio) !=3D 1) { > > - folio_unlock(folio); > > + arch_leave_lazy_mmu_mode(); > > + if (madvise_pte_split_folio(mm, pmd, addr= , > > + folio, &start= _pte, &ptl)) > > + nr =3D 0; > > + if (!start_pte) > > + break; > > + pte =3D start_pte; > > + arch_enter_lazy_mmu_mode(); > > continue; > > } > > + } > > > > + if (!folio_trylock(folio)) > > + continue; > > + /* > > + * If we have a large folio at this point, we know it is = fully mapped > > + * so if its mapcount is the same as its number of pages,= it must be > > + * exclusive. > > + */ > > + if (folio_mapcount(folio) !=3D folio_nr_pages(folio)) { > > + folio_unlock(folio); > > + continue; > > + } > > + if (folio_test_swapcache(folio) || folio_test_dirty(folio= )) { > > I don't understand the rationale for reducing the scope of this condition= al? > Previously it was used to avoid having to lock the folio if it wasn't in = the > swapcache or if it wasn't dirty. So now you will be locking much more oft= en. You're right. I should keep the previous behavior of avoiding locking the folio if wasn't in the swapcache or if it wasn't dirty. Thanks again for your time! Lance > > Thanks, > Ryan > > > if (folio_test_swapcache(folio) && > > !folio_free_swap(folio)) { > > folio_unlock(folio); > > continue; > > } > > - > > folio_clear_dirty(folio); > > - folio_unlock(folio); > > } > > + folio_unlock(folio); > > > > if (pte_young(ptent) || pte_dirty(ptent)) { > > - /* > > - * Some of architecture(ex, PPC) don't update TLB > > - * with set_pte_at and tlb_remove_tlb_entry so fo= r > > - * the portability, remap the pte with old|clean > > - * after pte clearing. > > - */ > > - ptent =3D ptep_get_and_clear_full(mm, addr, pte, > > - tlb->fullmm); > > - > > - ptent =3D pte_mkold(ptent); > > - ptent =3D pte_mkclean(ptent); > > - set_pte_at(mm, addr, pte, ptent); > > - tlb_remove_tlb_entry(tlb, pte, addr); > > + mkold_clean_ptes(vma, addr, pte, nr, tlb->fullmm)= ; > > + tlb_remove_tlb_entries(tlb, pte, nr, addr); > > } > > folio_mark_lazyfree(folio); > > } > > diff --git a/mm/memory.c b/mm/memory.c > > index 912cd738ec03..24769ecb59e5 100644 > > --- a/mm/memory.c > > +++ b/mm/memory.c > > @@ -989,7 +989,7 @@ copy_present_ptes(struct vm_area_struct *dst_vma, s= truct vm_area_struct *src_vma > > flags |=3D FPB_IGNORE_SOFT_DIRTY; > > > > nr =3D folio_pte_batch(folio, addr, src_pte, pte, max_nr,= flags, > > - &any_writable, NULL); > > + &any_writable, NULL, NULL); > > folio_ref_add(folio, nr); > > if (folio_test_anon(folio)) { > > if (unlikely(folio_try_dup_anon_rmap_ptes(folio, = page, > > @@ -1559,7 +1559,7 @@ static inline int zap_present_ptes(struct mmu_gat= her *tlb, > > */ > > if (unlikely(folio_test_large(folio) && max_nr !=3D 1)) { > > nr =3D folio_pte_batch(folio, addr, pte, ptent, max_nr, f= pb_flags, > > - NULL, NULL); > > + NULL, NULL, NULL); > > > > zap_present_folio_ptes(tlb, vma, folio, page, pte, ptent,= nr, > > addr, details, rss, force_flush, >