| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Tue, 28 Jun 2022 11:22:14 +0800
From: Huacai Chen <chenhuacai@...nel.org>
To: Muchun Song <songmuchun@...edance.com>
Cc: Huacai Chen <chenhuacai@...ngson.cn>,
Arnd Bergmann <arnd@...db.de>,
Thomas Bogendoerfer <tsbogend@...ha.franken.de>,
Dave Hansen <dave.hansen@...ux.intel.com>,
Andy Lutomirski <luto@...nel.org>,
Peter Zijlstra <peterz@...radead.org>,
Catalin Marinas <catalin.marinas@....com>,
Will Deacon <will@...nel.org>, loongarch@...ts.linux.dev,
linux-arch <linux-arch@...r.kernel.org>,
Xuefeng Li <lixuefeng@...ngson.cn>,
Guo Ren <guoren@...nel.org>, Xuerui Wang <kernel@...0n.name>,
Jiaxun Yang <jiaxun.yang@...goat.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Linux Memory Management List <linux-mm@...ck.org>,
"open list:MIPS" <linux-mips@...r.kernel.org>,
LKML <linux-kernel@...r.kernel.org>,
LAK <linux-arm-kernel@...ts.infradead.org>,
Feiyang Chen <chenfeiyang@...ngson.cn>,
Min Zhou <zhoumin@...ngson.cn>
Subject: Re: [PATCH 2/3] LoongArch: Add sparse memory vmemmap support
Hi, Muchun,
On Mon, Jun 27, 2022 at 6:33 PM Muchun Song <songmuchun@...edance.com> wrote:
>
> On Sat, Jun 25, 2022 at 5:54 PM Huacai Chen <chenhuacai@...ngson.cn> wrote:
> >
> > From: Feiyang Chen <chenfeiyang@...ngson.cn>
> >
> > Add sparse memory vmemmap support for LoongArch. SPARSEMEM_VMEMMAP
> > uses a virtually mapped memmap to optimise pfn_to_page and page_to_pfn
> > operations. This is the most efficient option when sufficient kernel
> > resources are available.
> >
> > Signed-off-by: Min Zhou <zhoumin@...ngson.cn>
> > Signed-off-by: Huacai Chen <chenhuacai@...ngson.cn>
> > Signed-off-by: Feiyang Chen <chenfeiyang@...ngson.cn>
> > ---
> > arch/loongarch/Kconfig | 2 +
> > arch/loongarch/include/asm/pgtable.h | 5 +-
> > arch/loongarch/include/asm/sparsemem.h | 8 +++
> > arch/loongarch/mm/init.c | 71 +++++++++++++++++++++++++-
> > include/linux/mm.h | 2 +
> > mm/sparse-vmemmap.c | 10 ++++
> > 6 files changed, 96 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
> > index dc19cf3071ea..8e56ca28165e 100644
> > --- a/arch/loongarch/Kconfig
> > +++ b/arch/loongarch/Kconfig
> > @@ -49,6 +49,7 @@ config LOONGARCH
> > select ARCH_USE_QUEUED_RWLOCKS
> > select ARCH_USE_QUEUED_SPINLOCKS
> > select ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT
> > + select ARCH_WANT_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
>
> I think this should be a separate patch to enable HVO (HugeTLB Vmemmap
> Optimization) since it is irrelevant to this patch.
It seems I have misunderstood HVO, then I will remove HVO parts from
this patch. Thank you.
Huacai
>
> Thanks.
>
> > select ARCH_WANTS_NO_INSTR
> > select BUILDTIME_TABLE_SORT
> > select COMMON_CLK
> > @@ -422,6 +423,7 @@ config ARCH_FLATMEM_ENABLE
> >
> > config ARCH_SPARSEMEM_ENABLE
> > def_bool y
> > + select SPARSEMEM_VMEMMAP_ENABLE
> > help
> > Say Y to support efficient handling of sparse physical memory,
> > for architectures which are either NUMA (Non-Uniform Memory Access)
> > diff --git a/arch/loongarch/include/asm/pgtable.h b/arch/loongarch/include/asm/pgtable.h
> > index 9c811c3f7572..b701ec7a0309 100644
> > --- a/arch/loongarch/include/asm/pgtable.h
> > +++ b/arch/loongarch/include/asm/pgtable.h
> > @@ -92,7 +92,10 @@ extern unsigned long zero_page_mask;
> > #define VMALLOC_START MODULES_END
> > #define VMALLOC_END \
> > (vm_map_base + \
> > - min(PTRS_PER_PGD * PTRS_PER_PUD * PTRS_PER_PMD * PTRS_PER_PTE * PAGE_SIZE, (1UL << cpu_vabits)) - PMD_SIZE)
> > + min(PTRS_PER_PGD * PTRS_PER_PUD * PTRS_PER_PMD * PTRS_PER_PTE * PAGE_SIZE, (1UL << cpu_vabits)) - PMD_SIZE - VMEMMAP_SIZE)
> > +
> > +#define vmemmap ((struct page *)((VMALLOC_END + PMD_SIZE) & PMD_MASK))
> > +#define VMEMMAP_END ((unsigned long)vmemmap + VMEMMAP_SIZE - 1)
> >
> > #define pte_ERROR(e) \
> > pr_err("%s:%d: bad pte %016lx.\n", __FILE__, __LINE__, pte_val(e))
> > diff --git a/arch/loongarch/include/asm/sparsemem.h b/arch/loongarch/include/asm/sparsemem.h
> > index 3d18cdf1b069..a1e440f6bec7 100644
> > --- a/arch/loongarch/include/asm/sparsemem.h
> > +++ b/arch/loongarch/include/asm/sparsemem.h
> > @@ -11,6 +11,14 @@
> > #define SECTION_SIZE_BITS 29 /* 2^29 = Largest Huge Page Size */
> > #define MAX_PHYSMEM_BITS 48
> >
> > +#ifndef CONFIG_SPARSEMEM_VMEMMAP
> > +#define VMEMMAP_SIZE 0
> > +#else
> > +#define VMEMMAP_SIZE (sizeof(struct page) * (1UL << (cpu_pabits + 1 - PAGE_SHIFT)))
> > +#endif
> > +
> > +#include <linux/mm_types.h>
> > +
> > #endif /* CONFIG_SPARSEMEM */
> >
> > #ifdef CONFIG_MEMORY_HOTPLUG
> > diff --git a/arch/loongarch/mm/init.c b/arch/loongarch/mm/init.c
> > index 7094a68c9b83..35128229fe46 100644
> > --- a/arch/loongarch/mm/init.c
> > +++ b/arch/loongarch/mm/init.c
> > @@ -22,7 +22,7 @@
> > #include <linux/pfn.h>
> > #include <linux/hardirq.h>
> > #include <linux/gfp.h>
> > -#include <linux/initrd.h>
> > +#include <linux/hugetlb.h>
> > #include <linux/mmzone.h>
> >
> > #include <asm/asm-offsets.h>
> > @@ -157,6 +157,75 @@ void arch_remove_memory(u64 start, u64 size, struct vmem_altmap *altmap)
> > #endif
> > #endif
> >
> > +#ifdef CONFIG_SPARSEMEM_VMEMMAP
> > +int __meminit vmemmap_populate_hugepages(unsigned long start, unsigned long end,
> > + int node, struct vmem_altmap *altmap)
> > +{
> > + unsigned long addr = start;
> > + unsigned long next;
> > + pgd_t *pgd;
> > + p4d_t *p4d;
> > + pud_t *pud;
> > + pmd_t *pmd;
> > +
> > + for (addr = start; addr < end; addr = next) {
> > + next = pmd_addr_end(addr, end);
> > +
> > + pgd = vmemmap_pgd_populate(addr, node);
> > + if (!pgd)
> > + return -ENOMEM;
> > + p4d = vmemmap_p4d_populate(pgd, addr, node);
> > + if (!p4d)
> > + return -ENOMEM;
> > + pud = vmemmap_pud_populate(p4d, addr, node);
> > + if (!pud)
> > + return -ENOMEM;
> > +
> > + pmd = pmd_offset(pud, addr);
> > + if (pmd_none(*pmd)) {
> > + void *p = NULL;
> > +
> > + p = vmemmap_alloc_block_buf(PMD_SIZE, node, NULL);
> > + if (p) {
> > + pmd_t entry;
> > +
> > + entry = pfn_pmd(virt_to_pfn(p), PAGE_KERNEL);
> > + pmd_val(entry) |= _PAGE_HUGE | _PAGE_HGLOBAL;
> > + set_pmd_at(&init_mm, addr, pmd, entry);
> > +
> > + continue;
> > + }
> > + } else if (pmd_val(*pmd) & _PAGE_HUGE) {
> > + vmemmap_verify((pte_t *)pmd, node, addr, next);
> > + continue;
> > + }
> > + if (vmemmap_populate_basepages(addr, next, node, NULL))
> > + return -ENOMEM;
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +#if CONFIG_PGTABLE_LEVELS == 2
> > +int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
> > + struct vmem_altmap *altmap)
> > +{
> > + return vmemmap_populate_basepages(start, end, node, NULL);
> > +}
> > +#else
> > +int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
> > + struct vmem_altmap *altmap)
> > +{
> > + return vmemmap_populate_hugepages(start, end, node, NULL);
> > +}
> > +#endif
> > +
> > +void vmemmap_free(unsigned long start, unsigned long end,
> > + struct vmem_altmap *altmap)
> > +{
> > +}
> > +#endif
> > +
> > /*
> > * Align swapper_pg_dir in to 64K, allows its address to be loaded
> > * with a single LUI instruction in the TLB handlers. If we used
> > diff --git a/include/linux/mm.h b/include/linux/mm.h
> > index bc8f326be0ce..3472b924a1ea 100644
> > --- a/include/linux/mm.h
> > +++ b/include/linux/mm.h
> > @@ -3203,6 +3203,8 @@ void *sparse_buffer_alloc(unsigned long size);
> > struct page * __populate_section_memmap(unsigned long pfn,
> > unsigned long nr_pages, int nid, struct vmem_altmap *altmap,
> > struct dev_pagemap *pgmap);
> > +void pmd_init(void *addr);
> > +void pud_init(void *addr);
> > pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
> > p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node);
> > pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node);
> > diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
> > index f4fa61dbbee3..33e2a1ceee72 100644
> > --- a/mm/sparse-vmemmap.c
> > +++ b/mm/sparse-vmemmap.c
> > @@ -587,6 +587,10 @@ pmd_t * __meminit vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node)
> > return pmd;
> > }
> >
> > +void __weak __meminit pmd_init(void *addr)
> > +{
> > +}
> > +
> > pud_t * __meminit vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node)
> > {
> > pud_t *pud = pud_offset(p4d, addr);
> > @@ -594,11 +598,16 @@ pud_t * __meminit vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node)
> > void *p = vmemmap_alloc_block_zero(PAGE_SIZE, node);
> > if (!p)
> > return NULL;
> > + pmd_init(p);
> > pud_populate(&init_mm, pud, p);
> > }
> > return pud;
> > }
> >
> > +void __weak __meminit pud_init(void *addr)
> > +{
> > +}
> > +
> > p4d_t * __meminit vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node)
> > {
> > p4d_t *p4d = p4d_offset(pgd, addr);
> > @@ -606,6 +615,7 @@ p4d_t * __meminit vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node)
> > void *p = vmemmap_alloc_block_zero(PAGE_SIZE, node);
> > if (!p)
> > return NULL;
> > + pud_init(p);
> > p4d_populate(&init_mm, p4d, p);
> > }
> > return p4d;
> > --
> > 2.27.0
> >
Powered by blists - more mailing lists