[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAEn-LTo=GP5OMZiaBi8BL1etLcGrCyofQrtQ4-JOo5zcpCLu8A@mail.gmail.com>
Date: Sat, 21 Mar 2020 22:29:06 +0200
From: David Abdurachmanov <david.abdurachmanov@...il.com>
To: Palmer Dabbelt <palmerdabbelt@...gle.com>
Cc: Atish Patra <Atish.Patra@....com>,
"linux-kernel@...r.kernel.org List" <linux-kernel@...r.kernel.org>,
Albert Ou <aou@...s.berkeley.edu>, nickhu@...estech.com,
Bjorn Topel <bjorn.topel@...il.com>,
Anup Patel <Anup.Patel@....com>, stable@...r.kernel.org,
Mike Rapoport <rppt@...ux.ibm.com>,
Thomas Gleixner <tglx@...utronix.de>,
Paul Walmsley <paul.walmsley@...ive.com>,
Greentime Hu <greentime.hu@...ive.com>,
akpm@...ux-foundation.org,
linux-riscv <linux-riscv@...ts.infradead.org>
Subject: Re: [PATCH] RISC-V: Move all address space definition macros to one place
On Fri, Mar 6, 2020 at 2:20 AM Palmer Dabbelt <palmerdabbelt@...gle.com> wrote:
>
> On Mon, 24 Feb 2020 11:34:36 PST (-0800), Atish Patra wrote:
> > If both CONFIG_KASAN and CONFIG_SPARSEMEM_VMEMMAP are set, we get the
> > following compilation error.
> >
> > ---------------------------------------------------------------
> > ./arch/riscv/include/asm/pgtable-64.h: In function ‘pud_page’:
> > ./include/asm-generic/memory_model.h:54:29: error: ‘vmemmap’ undeclared
> > (first use in this function); did you mean ‘mem_map’?
> > #define __pfn_to_page(pfn) (vmemmap + (pfn))
> > ^~~~~~~
> > ./include/asm-generic/memory_model.h:82:21: note: in expansion of
> > macro ‘__pfn_to_page’
> >
> > #define pfn_to_page __pfn_to_page
> > ^~~~~~~~~~~~~
> > ./arch/riscv/include/asm/pgtable-64.h:70:9: note: in expansion of macro
> > ‘pfn_to_page’
> > return pfn_to_page(pud_val(pud) >> _PAGE_PFN_SHIFT);
> > ---------------------------------------------------------------
> >
> > Fix the compliation errors by moving all the address space definition
> > macros before including pgtable-64.h.
> >
> > Cc: stable@...r.kernel.org
> > Fixes: 8ad8b72721d0 (riscv: Add KASAN support)
> >
> > Signed-off-by: Atish Patra <atish.patra@....com>
> > ---
> > arch/riscv/include/asm/pgtable.h | 78 +++++++++++++++++---------------
> > 1 file changed, 41 insertions(+), 37 deletions(-)
> >
> > diff --git a/arch/riscv/include/asm/pgtable.h b/arch/riscv/include/asm/pgtable.h
> > index 453afb0a570a..4f6ee48a42e8 100644
> > --- a/arch/riscv/include/asm/pgtable.h
> > +++ b/arch/riscv/include/asm/pgtable.h
> > @@ -19,6 +19,47 @@
> > #include <asm/tlbflush.h>
> > #include <linux/mm_types.h>
> >
> > +#ifdef CONFIG_MMU
> > +
> > +#define VMALLOC_SIZE (KERN_VIRT_SIZE >> 1)
> > +#define VMALLOC_END (PAGE_OFFSET - 1)
> > +#define VMALLOC_START (PAGE_OFFSET - VMALLOC_SIZE)
> > +
> > +#define BPF_JIT_REGION_SIZE (SZ_128M)
> > +#define BPF_JIT_REGION_START (PAGE_OFFSET - BPF_JIT_REGION_SIZE)
> > +#define BPF_JIT_REGION_END (VMALLOC_END)
> > +
> > +/*
> > + * Roughly size the vmemmap space to be large enough to fit enough
> > + * struct pages to map half the virtual address space. Then
> > + * position vmemmap directly below the VMALLOC region.
> > + */
> > +#define VMEMMAP_SHIFT \
> > + (CONFIG_VA_BITS - PAGE_SHIFT - 1 + STRUCT_PAGE_MAX_SHIFT)
> > +#define VMEMMAP_SIZE BIT(VMEMMAP_SHIFT)
> > +#define VMEMMAP_END (VMALLOC_START - 1)
> > +#define VMEMMAP_START (VMALLOC_START - VMEMMAP_SIZE)
> > +
> > +/*
> > + * Define vmemmap for pfn_to_page & page_to_pfn calls. Needed if kernel
> > + * is configured with CONFIG_SPARSEMEM_VMEMMAP enabled.
> > + */
> > +#define vmemmap ((struct page *)VMEMMAP_START)
> > +
> > +#define PCI_IO_SIZE SZ_16M
> > +#define PCI_IO_END VMEMMAP_START
> > +#define PCI_IO_START (PCI_IO_END - PCI_IO_SIZE)
> > +
> > +#define FIXADDR_TOP PCI_IO_START
> > +#ifdef CONFIG_64BIT
> > +#define FIXADDR_SIZE PMD_SIZE
> > +#else
> > +#define FIXADDR_SIZE PGDIR_SIZE
> > +#endif
> > +#define FIXADDR_START (FIXADDR_TOP - FIXADDR_SIZE)
> > +
> > +#endif
> > +
> > #ifdef CONFIG_64BIT
> > #include <asm/pgtable-64.h>
> > #else
> > @@ -90,31 +131,6 @@ extern pgd_t swapper_pg_dir[];
> > #define __S110 PAGE_SHARED_EXEC
> > #define __S111 PAGE_SHARED_EXEC
> >
> > -#define VMALLOC_SIZE (KERN_VIRT_SIZE >> 1)
> > -#define VMALLOC_END (PAGE_OFFSET - 1)
> > -#define VMALLOC_START (PAGE_OFFSET - VMALLOC_SIZE)
> > -
> > -#define BPF_JIT_REGION_SIZE (SZ_128M)
> > -#define BPF_JIT_REGION_START (PAGE_OFFSET - BPF_JIT_REGION_SIZE)
> > -#define BPF_JIT_REGION_END (VMALLOC_END)
> > -
> > -/*
> > - * Roughly size the vmemmap space to be large enough to fit enough
> > - * struct pages to map half the virtual address space. Then
> > - * position vmemmap directly below the VMALLOC region.
> > - */
> > -#define VMEMMAP_SHIFT \
> > - (CONFIG_VA_BITS - PAGE_SHIFT - 1 + STRUCT_PAGE_MAX_SHIFT)
> > -#define VMEMMAP_SIZE BIT(VMEMMAP_SHIFT)
> > -#define VMEMMAP_END (VMALLOC_START - 1)
> > -#define VMEMMAP_START (VMALLOC_START - VMEMMAP_SIZE)
> > -
> > -/*
> > - * Define vmemmap for pfn_to_page & page_to_pfn calls. Needed if kernel
> > - * is configured with CONFIG_SPARSEMEM_VMEMMAP enabled.
> > - */
> > -#define vmemmap ((struct page *)VMEMMAP_START)
> > -
> > static inline int pmd_present(pmd_t pmd)
> > {
> > return (pmd_val(pmd) & (_PAGE_PRESENT | _PAGE_PROT_NONE));
> > @@ -452,18 +468,6 @@ static inline int ptep_clear_flush_young(struct vm_area_struct *vma,
> > #define __pte_to_swp_entry(pte) ((swp_entry_t) { pte_val(pte) })
> > #define __swp_entry_to_pte(x) ((pte_t) { (x).val })
> >
> > -#define PCI_IO_SIZE SZ_16M
> > -#define PCI_IO_END VMEMMAP_START
> > -#define PCI_IO_START (PCI_IO_END - PCI_IO_SIZE)
> > -
> > -#define FIXADDR_TOP PCI_IO_START
> > -#ifdef CONFIG_64BIT
> > -#define FIXADDR_SIZE PMD_SIZE
> > -#else
> > -#define FIXADDR_SIZE PGDIR_SIZE
> > -#endif
> > -#define FIXADDR_START (FIXADDR_TOP - FIXADDR_SIZE)
> > -
> > /*
> > * Task size is 0x4000000000 for RV64 or 0x9fc00000 for RV32.
> > * Note that PGDIR_SIZE must evenly divide TASK_SIZE.
>
> While this isn't technically a fix, I'm inclined to target it for the RCs just
> to avoid conflicts. I've put it on for-next now so the builders have some time
> to chew on things, as I don't want to put in a non-fix too quickly.
I hit the same issue in Fedora/RISCV while building kernel-5.6.0-0.rc6, and
we don't have KASAN selected. We do have CONFIG_SPARSEMEM_VMEMMAP
selected.
I will try this patch tomorrow.
>
> Thanks!
Powered by blists - more mailing lists