From: Huacai Chen <chenhuacai@kernel.org>
To: bibo mao <maobibo@loongson.cn>
Cc: Dennis Zhou <dennis@kernel.org>, Tejun Heo <tj@kernel.org>,
Christoph Lameter <cl@linux.com>,
Andrew Morton <akpm@linux-foundation.org>,
loongarch@lists.linux.dev, linux-kernel@vger.kernel.org,
linux-mm@kvack.org, WANG Xuerui <kernel@xen0n.name>
Subject: Re: [PATCH 3/3] LoongArch: mm: Add unified function populate_kernel_pte
Date: Thu, 10 Aug 2023 12:27:11 +0800 [thread overview]
Message-ID: <CAAhV-H7kBP-qME2mxwQKqYeggWu4ApqAJukz3eZxfqsYzRvGEA@mail.gmail.com> (raw)
In-Reply-To: <b4b21efb-a0c7-4625-91e7-4d0c5b80a249@loongson.cn>
On Thu, Aug 10, 2023 at 12:09 PM bibo mao <maobibo@loongson.cn> wrote:
>
>
>
> 在 2023/8/2 15:25, Huacai Chen 写道:
> > On Tue, Aug 1, 2023 at 9:22 AM bibo mao <maobibo@loongson.cn> wrote:
> >>
> >>
> >>
> >> 在 2023/7/31 22:15, Huacai Chen 写道:
> >>> On Wed, Jul 12, 2023 at 11:16 AM Bibo Mao <maobibo@loongson.cn> wrote:
> >>>>
> >>>> Function pcpu_populate_pte and fixmap_pte are similar, they populate
> >>>> one page from kernel address space. And there is confusion between
> >>>> pgd and p4d in function fixmap_pte, such as pgd_none always returns
> >>>> zero. This patch adds unified function populate_kernel_pte and replaces
> >>>> pcpu_populate_pte and fixmap_pte.
> >>>>
> >>>> Signed-off-by: Bibo Mao <maobibo@loongson.cn>
> >>>> ---
> >>>> arch/loongarch/include/asm/pgalloc.h | 1 +
> >>>> arch/loongarch/kernel/numa.c | 40 +--------------------
> >>>> arch/loongarch/mm/init.c | 52 ++++++++++++++++------------
> >>>> 3 files changed, 32 insertions(+), 61 deletions(-)
> >>>>
> >>>> diff --git a/arch/loongarch/include/asm/pgalloc.h b/arch/loongarch/include/asm/pgalloc.h
> >>>> index af1d1e4a6965..ca17b573dba6 100644
> >>>> --- a/arch/loongarch/include/asm/pgalloc.h
> >>>> +++ b/arch/loongarch/include/asm/pgalloc.h
> >>>> @@ -91,4 +91,5 @@ static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long address)
> >>>>
> >>>> #endif /* __PAGETABLE_PUD_FOLDED */
> >>>>
> >>>> +extern pte_t * __init populate_kernel_pte(unsigned long addr);
> >>>> #endif /* _ASM_PGALLOC_H */
> >>>> diff --git a/arch/loongarch/kernel/numa.c b/arch/loongarch/kernel/numa.c
> >>>> index 778e1c20bfb0..24a693b76873 100644
> >>>> --- a/arch/loongarch/kernel/numa.c
> >>>> +++ b/arch/loongarch/kernel/numa.c
> >>>> @@ -67,46 +67,8 @@ static int __init pcpu_cpu_distance(unsigned int from, unsigned int to)
> >>>>
> >>>> void __init pcpu_populate_pte(unsigned long addr)
> >>>> {
> >>>> - pgd_t *pgd = pgd_offset_k(addr);
> >>>> - p4d_t *p4d = p4d_offset(pgd, addr);
> >>>> - pud_t *pud;
> >>>> - pmd_t *pmd;
> >>>> -
> >>>> - if (p4d_none(*p4d)) {
> >>>> - pud = memblock_alloc_raw(PAGE_SIZE, PAGE_SIZE);
> >>>> - if (!pud)
> >>>> - goto err_alloc;
> >>>> - p4d_populate(&init_mm, p4d, pud);
> >>>> -#ifndef __PAGETABLE_PUD_FOLDED
> >>>> - pud_init(pud);
> >>>> -#endif
> >>>> - }
> >>>> -
> >>>> - pud = pud_offset(p4d, addr);
> >>>> - if (pud_none(*pud)) {
> >>>> - pmd = memblock_alloc_raw(PAGE_SIZE, PAGE_SIZE);
> >>>> - if (!pmd)
> >>>> - goto err_alloc;
> >>>> - pud_populate(&init_mm, pud, pmd);
> >>>> -#ifndef __PAGETABLE_PMD_FOLDED
> >>>> - pmd_init(pmd);
> >>>> -#endif
> >>>> - }
> >>>> -
> >>>> - pmd = pmd_offset(pud, addr);
> >>>> - if (!pmd_present(*pmd)) {
> >>>> - pte_t *pte;
> >>>> -
> >>>> - pte = memblock_alloc(PAGE_SIZE, PAGE_SIZE);
> >>>> - if (!pte)
> >>>> - goto err_alloc;
> >>>> - pmd_populate_kernel(&init_mm, pmd, pte);
> >>>> - }
> >>>> -
> >>>> + populate_kernel_pte(addr);
> >>>> return;
> >>>> -
> >>>> -err_alloc:
> >>>> - panic("%s: Failed to allocate memory\n", __func__);
> >>>> }
> >>>>
> >>>> void __init setup_per_cpu_areas(void)
> >>>> diff --git a/arch/loongarch/mm/init.c b/arch/loongarch/mm/init.c
> >>>> index 3b7d8129570b..6cd2948373ae 100644
> >>>> --- a/arch/loongarch/mm/init.c
> >>>> +++ b/arch/loongarch/mm/init.c
> >>>> @@ -191,46 +191,49 @@ void vmemmap_free(unsigned long start, unsigned long end, struct vmem_altmap *al
> >>>> #endif
> >>>> #endif
> >>>>
> >>>> -static pte_t *fixmap_pte(unsigned long addr)
> >>>> +pte_t * __init populate_kernel_pte(unsigned long addr)
> >>>> {
> >>>> - pgd_t *pgd;
> >>>> - p4d_t *p4d;
> >>>> + pgd_t *pgd = pgd_offset_k(addr);
> >>>> + p4d_t *p4d = p4d_offset(pgd, addr);
> >>>> pud_t *pud;
> >>>> pmd_t *pmd;
> >>>>
> >>>> - pgd = pgd_offset_k(addr);
> >>>> - p4d = p4d_offset(pgd, addr);
> >>>> -
> >>>> - if (pgd_none(*pgd)) {
> >>>> - pud_t *new __maybe_unused;
> >>>> -
> >>>> - new = memblock_alloc_low(PAGE_SIZE, PAGE_SIZE);
> >>>> - pgd_populate(&init_mm, pgd, new);
> >>>> + if (p4d_none(*p4d)) {
> >>>> + pud = memblock_alloc_raw(PAGE_SIZE, PAGE_SIZE);
> >>>> + if (!pud)
> >>>> + goto err_alloc;
> >>>> + p4d_populate(&init_mm, p4d, pud);
> >>>> #ifndef __PAGETABLE_PUD_FOLDED
> >>>> - pud_init(new);
> >>>> + pud_init(pud);
> >>>> #endif
> >>>> }
> >>>>
> >>>> pud = pud_offset(p4d, addr);
> >>>> if (pud_none(*pud)) {
> >>>> - pmd_t *new __maybe_unused;
> >>>> -
> >>>> - new = memblock_alloc_low(PAGE_SIZE, PAGE_SIZE);
> >>>> - pud_populate(&init_mm, pud, new);
> >>>> + pmd = memblock_alloc_raw(PAGE_SIZE, PAGE_SIZE);
> >>>> + if (!pmd)
> >>>> + goto err_alloc;
> >>>> + pud_populate(&init_mm, pud, pmd);
> >>>> #ifndef __PAGETABLE_PMD_FOLDED
> >>>> - pmd_init(new);
> >>>> + pmd_init(pmd);
> >>>> #endif
> >>>> }
> >>>>
> >>>> pmd = pmd_offset(pud, addr);
> >>>> - if (pmd_none(*pmd)) {
> >>>> - pte_t *new __maybe_unused;
> >>>> + if (!pmd_present(*pmd)) {
> >>>> + pte_t *pte;
> >>>>
> >>>> - new = memblock_alloc_low(PAGE_SIZE, PAGE_SIZE);
> >>>> - pmd_populate_kernel(&init_mm, pmd, new);
> >>>> + pte = memblock_alloc(PAGE_SIZE, PAGE_SIZE);
> >>> I don't think memblock_alloc_low() here can be replaced by memblock_alloc().
> >> Can you share me the points that pte table must be allocated with memblock_alloc_low
> >> in this place?
> > I forget the reason now, so if you confirm memblock_alloc() works well
> > here, you can use it. But please don't use memblock_alloc_raw().
> what a mess, there is more comments if there is special reason, else everyone can
> forgot by elapsed time.
>
> why the function memblock_alloc_raw can not be use? there is one useless page copy.
This is not a performance critical path, keeping consistency with
mm/percpu.c can make life easier.
Huacai
>
> Regards
> Bibo Mao
>
>
> >
> > Huacai
> >>
> >> Regards
> >> Bibo Mao
> >>>
> >>>
> >>> Huacai
> >>>> + if (!pte)
> >>>> + goto err_alloc;
> >>>> + pmd_populate_kernel(&init_mm, pmd, pte);
> >>>> }
> >>>>
> >>>> return pte_offset_kernel(pmd, addr);
> >>>> +
> >>>> +err_alloc:
> >>>> + panic("%s: Failed to allocate memory\n", __func__);
> >>>> + return NULL;
> >>>> }
> >>>>
> >>>> void __init __set_fixmap(enum fixed_addresses idx,
> >>>> @@ -241,7 +244,12 @@ void __init __set_fixmap(enum fixed_addresses idx,
> >>>>
> >>>> BUG_ON(idx <= FIX_HOLE || idx >= __end_of_fixed_addresses);
> >>>>
> >>>> - ptep = fixmap_pte(addr);
> >>>> + /*
> >>>> + * Now only FIX_EARLYCON_MEM_BASE fixed map is used
> >>>> + * __set_fixmap must be called before mem_init since function
> >>>> + * populate_kernel_pte allocates memory with memblock_alloc method.
> >>>> + */
> >>>> + ptep = populate_kernel_pte(addr);
> >>>> if (!pte_none(*ptep)) {
> >>>> pte_ERROR(*ptep);
> >>>> return;
> >>>> --
> >>>> 2.27.0
> >>>>
> >>
> >>
>
>
next prev parent reply other threads:[~2023-08-10 4:27 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-07-12 3:16 [PATCH 0/3] LoongArch: mm: Code cleanup with populate pte Bibo Mao
2023-07-12 3:16 ` [PATCH 1/3] mm/percpu: Remove some local variables in pcpu_populate_pte Bibo Mao
2023-07-27 23:08 ` Dennis Zhou
2023-07-28 1:13 ` bibo mao
2023-07-12 3:16 ` [PATCH 2/3] LoongArch: Code cleanup in function pcpu_populate_pte Bibo Mao
2023-07-31 14:15 ` Huacai Chen
2023-08-01 1:18 ` bibo mao
2023-07-12 3:16 ` [PATCH 3/3] LoongArch: mm: Add unified function populate_kernel_pte Bibo Mao
2023-07-31 14:15 ` Huacai Chen
2023-08-01 1:22 ` bibo mao
2023-08-02 7:25 ` Huacai Chen
2023-08-10 4:08 ` bibo mao
2023-08-10 4:27 ` Huacai Chen [this message]
2023-08-10 4:42 ` bibo mao
2023-07-25 0:36 ` [PATCH 0/3] LoongArch: mm: Code cleanup with populate pte bibo mao
2023-07-25 2:33 ` Dennis Zhou
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=CAAhV-H7kBP-qME2mxwQKqYeggWu4ApqAJukz3eZxfqsYzRvGEA@mail.gmail.com \
--to=chenhuacai@kernel.org \
--cc=akpm@linux-foundation.org \
--cc=cl@linux.com \
--cc=dennis@kernel.org \
--cc=kernel@xen0n.name \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=loongarch@lists.linux.dev \
--cc=maobibo@loongson.cn \
--cc=tj@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
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