Hi Kazuhito, Looks good -- queued for crash-7.2.2:
https://github.com/crash-utility/crash/commit/4141373d9de3fea29f5d2b58f60e44bc132726c0 Thanks, Dave ----- Original Message ----- > changes v2 -> v3: > - move the setting point of machdep->is_page_ptr to machdep_init(SETUP_ENV) > changes v1 -> v2: > - rewrite based on the per-architecture function call Dave provided > - remove the part which used page.flags for non-VMEMMAP kernels > - add address range/position check first > - remove/optimize the calculations of mem_map and phys address > - modify the patch description > > The "kmem -[sS]" commands can take several minutes to complete with > the following conditions: > - The system has a lot of memory sections with CONFIG_SPARSEMEM, and > - The kernel uses SLUB and it has a very long partial slab list. > > crash> kmem -s dentry > CACHE NAME OBJSIZE ALLOCATED TOTAL SLABS > SSIZE > ffff88017fc78a00 dentry 192 9038949 10045728 239184 > 8k > crash> kmem -s dentry | bash -c 'cat >/dev/null ; echo $SECONDS' > 133 > crash> kmem -S dentry | bash -c 'cat >/dev/null ; echo $SECONDS' > 656 > > One of the causes is that is_page_ptr() in count_partial() determines > whether a given slub page address is a page struct by searching all > mem_sections available for the one which includes it. > > With CONFIG_SPARSEMEM_VMEMMAP on x86_64, we can do that by checking > its address range and whether its calculated mem_section is valid. > With this patch, the computation amount can be significantly reduced > in that case. > > crash> kmem -s dentry | bash -c 'cat >/dev/null ; echo $SECONDS' > 1 > crash> kmem -S dentry | bash -c 'cat >/dev/null ; echo $SECONDS' > 1 > > Signed-off-by: Kazuhito Hagio <k-ha...@ab.jp.nec.com> > --- > defs.h | 1 + > x86_64.c | 23 +++++++++++++++++++++++ > 2 files changed, 24 insertions(+) > > diff --git a/defs.h b/defs.h > index 9663bd8..7998ebf 100644 > --- a/defs.h > +++ b/defs.h > @@ -5133,6 +5133,7 @@ int vaddr_type(ulong, struct task_context *); > char *format_stack_entry(struct bt_info *bt, char *, ulong, ulong); > int in_user_stack(ulong, ulong); > int dump_inode_page(ulong); > +ulong valid_section_nr(ulong); > > > /* > diff --git a/x86_64.c b/x86_64.c > index 7449571..0d5e150 100644 > --- a/x86_64.c > +++ b/x86_64.c > @@ -77,6 +77,7 @@ static void x86_64_calc_phys_base(void); > static int x86_64_is_module_addr(ulong); > static int x86_64_is_kvaddr(ulong); > static int x86_64_is_uvaddr(ulong, struct task_context *); > +static int x86_64_is_page_ptr(ulong, physaddr_t *); > static ulong *x86_64_kpgd_offset(ulong, int, int); > static ulong x86_64_upgd_offset(struct task_context *, ulong, int, int); > static ulong x86_64_upgd_offset_legacy(struct task_context *, ulong, int, > int); > @@ -156,6 +157,7 @@ x86_64_init(int when) > { > case SETUP_ENV: > machdep->process_elf_notes = x86_process_elf_notes; > + machdep->is_page_ptr = x86_64_is_page_ptr; > break; > case PRE_SYMTAB: > machdep->verify_symbol = x86_64_verify_symbol; > @@ -802,6 +804,7 @@ x86_64_dump_machdep_table(ulong arg) > fprintf(fp, " get_smp_cpus: x86_64_get_smp_cpus()\n"); > fprintf(fp, " is_kvaddr: x86_64_is_kvaddr()\n"); > fprintf(fp, " is_uvaddr: x86_64_is_uvaddr()\n"); > + fprintf(fp, " is_page_ptr: x86_64_is_page_ptr()\n"); > fprintf(fp, " verify_paddr: x86_64_verify_paddr()\n"); > fprintf(fp, " get_kvaddr_ranges: x86_64_get_kvaddr_ranges()\n"); > fprintf(fp, " init_kernel_pgd: x86_64_init_kernel_pgd()\n"); > @@ -1594,6 +1597,26 @@ x86_64_is_uvaddr(ulong addr, struct task_context *tc) > return (addr < USERSPACE_TOP); > } > > +static int > +x86_64_is_page_ptr(ulong addr, physaddr_t *phys) > +{ > + ulong pfn, nr; > + > + if (IS_SPARSEMEM() && (machdep->flags & VMEMMAP) && > + (addr >= VMEMMAP_VADDR && addr <= VMEMMAP_END) && > + !((addr - VMEMMAP_VADDR) % SIZE(page))) { > + > + pfn = (addr - VMEMMAP_VADDR) / SIZE(page); > + nr = pfn_to_section_nr(pfn); > + if (valid_section_nr(nr)) { > + if (phys) > + *phys = PTOB(pfn); > + return TRUE; > + } > + } > + return FALSE; > +} > + > /* > * Find the kernel pgd entry.. > * pgd = pgd_offset_k(addr); > -- > 1.8.3.1 > > -- > Crash-utility mailing list > Crash-utility@redhat.com > https://www.redhat.com/mailman/listinfo/crash-utility > -- Crash-utility mailing list Crash-utility@redhat.com https://www.redhat.com/mailman/listinfo/crash-utility
