[RFC PATCH 4/4] arm64: support flipped VA and 52-bit kernel VA
HAGIO KAZUHITO(萩尾 一仁)
k-hagio-ab at nec.com
Thu Jan 14 19:40:08 EST 2021
Hi Pingfan, Bhupesh,
Thank you for your comments!
-----Original Message-----
> Hi Kazu,
>
> On Thu, Jan 14, 2021 at 3:33 PM piliu <piliu at redhat.com> wrote:
> >
> >
> >
> > On 1/14/21 4:25 PM, kazuhito.hagio at gmail.com wrote:
> > > From: Kazuhito Hagio <k-hagio-ab at nec.com>
> > >
> > > Based on Bhupesh's patch and contains Pingfan's idea.
> > >
> > > Signed-off-by: Bhupesh Sharma <bhsharma at redhat.com>
> > > Signed-off-by: Kazuhito Hagio <k-hagio-ab at nec.com>
> > > ---
> > > arch/arm64.c | 95 ++++++++++++++++++++++++++++++++++++++++++++++++----------
> > > makedumpfile.c | 2 ++
> > > makedumpfile.h | 1 +
> > > 3 files changed, 83 insertions(+), 15 deletions(-)
> > >
> > > diff --git a/arch/arm64.c b/arch/arm64.c
> > > index 61ec89a..4ece19d 100644
> > > --- a/arch/arm64.c
> > > +++ b/arch/arm64.c
> > > @@ -47,6 +47,8 @@ typedef struct {
> > > static int lpa_52_bit_support_available;
> > > static int pgtable_level;
> > > static int va_bits;
> > > +static int vabits_actual;
> > > +static int flipped_va;
> > > static unsigned long kimage_voffset;
> > >
> > > #define SZ_4K 4096
> > > @@ -58,7 +60,6 @@ static unsigned long kimage_voffset;
> > > #define PAGE_OFFSET_42 ((0xffffffffffffffffUL) << 42)
> > > #define PAGE_OFFSET_47 ((0xffffffffffffffffUL) << 47)
> > > #define PAGE_OFFSET_48 ((0xffffffffffffffffUL) << 48)
> > > -#define PAGE_OFFSET_52 ((0xffffffffffffffffUL) << 52)
> > >
> > > #define pgd_val(x) ((x).pgd)
> > > #define pud_val(x) (pgd_val((x).pgd))
> > > @@ -218,12 +219,20 @@ pmd_page_paddr(pmd_t pmd)
> > > #define pte_index(vaddr) (((vaddr) >> PAGESHIFT()) & (PTRS_PER_PTE - 1))
> > > #define pte_offset(dir, vaddr) (pmd_page_paddr((*dir)) + pte_index(vaddr) *
> sizeof(pte_t))
> > >
> > > +/*
> > > + * The linear kernel range starts at the bottom of the virtual address
> > > + * space. Testing the top bit for the start of the region is a
> > > + * sufficient check and avoids having to worry about the tag.
> > > + */
> > > +#define is_linear_addr(addr) (flipped_va ? \
> > > + (!((unsigned long)(addr) & (1UL << (vabits_actual - 1)))) : \
> > > + (!!((unsigned long)(addr) & (1UL << (vabits_actual - 1)))))
> > > +
> > > static unsigned long long
> > > __pa(unsigned long vaddr)
> > > {
> > > - if (kimage_voffset == NOT_FOUND_NUMBER ||
> > > - (vaddr >= PAGE_OFFSET))
> > > - return (vaddr - PAGE_OFFSET + info->phys_base);
> > > + if (kimage_voffset == NOT_FOUND_NUMBER || is_linear_addr(vaddr))
> > > + return ((vaddr & ~PAGE_OFFSET) + info->phys_base);
> > > else
> > > return (vaddr - kimage_voffset);
> > > }
> > > @@ -253,6 +262,7 @@ static int calculate_plat_config(void)
> > > (PAGESIZE() == SZ_64K && va_bits == 42)) {
> > > pgtable_level = 2;
> > > } else if ((PAGESIZE() == SZ_64K && va_bits == 48) ||
> > > + (PAGESIZE() == SZ_64K && va_bits == 52) ||
> > > (PAGESIZE() == SZ_4K && va_bits == 39) ||
> > > (PAGESIZE() == SZ_16K && va_bits == 47)) {
> > > pgtable_level = 3;
> > > @@ -263,6 +273,7 @@ static int calculate_plat_config(void)
> > > PAGESIZE(), va_bits);
> > > return FALSE;
> > > }
> > > + DEBUG_MSG("pgtable_level: %d\n", pgtable_level);
> > >
> > > return TRUE;
> > > }
> > > @@ -383,22 +394,54 @@ get_va_bits_from_stext_arm64(void)
> > > return TRUE;
> > > }
> > >
> > > +static void
> > > +get_page_offset_arm64(void)
> > > +{
> > > + ulong page_end;
> > > + int vabits_min;
> > > +
> > > + /*
> > > + * See arch/arm64/include/asm/memory.h for more details of
> > > + * the PAGE_OFFSET calculation.
> > > + */
> > > + vabits_min = (va_bits > 48) ? 48 : va_bits;
> > > + page_end = -(1UL << (vabits_min - 1));
> > > +
> > > + if (SYMBOL(_stext) > page_end) {
> > > + flipped_va = TRUE;
> > > + info->page_offset = -(1UL << vabits_actual);
> > > + } else {
> > > + flipped_va = FALSE;
> > > + info->page_offset = -(1UL << (vabits_actual - 1));
> > > + }
> > > +
> > > + DEBUG_MSG("page_offset : %lx (from page_end check)\n",
> > > + info->page_offset);
> > > +}
> > > +
> > > int
> > > get_machdep_info_arm64(void)
> > > {
> > > + /* Check if va_bits is still not initialized. If still 0, call
> > > + * get_versiondep_info() to initialize the same.
> > > + */
> > > + if (!va_bits)
> > > + get_versiondep_info_arm64();
> > > +
> > > /* Determine if the PA address range is 52-bits: ARMv8.2-LPA */
> > > if (NUMBER(MAX_PHYSMEM_BITS) != NOT_FOUND_NUMBER) {
> > > info->max_physmem_bits = NUMBER(MAX_PHYSMEM_BITS);
> > > + DEBUG_MSG("max_physmem_bits : %ld (vmcoreinfo)\n", info->max_physmem_bits);
> > > if (info->max_physmem_bits == 52)
> > > lpa_52_bit_support_available = 1;
> > > - } else
> > > - info->max_physmem_bits = 48;
> > > + } else {
> > > + if (va_bits == 52)
> > > + info->max_physmem_bits = 52; /* just guess */
> > > + else
> > > + info->max_physmem_bits = 48;
> > >
> > > - /* Check if va_bits is still not initialized. If still 0, call
> > > - * get_versiondep_info() to initialize the same.
> > > - */
> > > - if (!va_bits)
> > > - get_versiondep_info_arm64();
> > > + DEBUG_MSG("max_physmem_bits : %ld (guess)\n", info->max_physmem_bits);
> > > + }
> > >
> > > if (!calculate_plat_config()) {
> > > ERRMSG("Can't determine platform config values\n");
> > > @@ -409,7 +452,6 @@ get_machdep_info_arm64(void)
> > > info->section_size_bits = SECTIONS_SIZE_BITS;
> > >
> > > DEBUG_MSG("kimage_voffset : %lx\n", kimage_voffset);
> > > - DEBUG_MSG("max_physmem_bits : %ld\n", info->max_physmem_bits);
> > > DEBUG_MSG("section_size_bits: %ld\n", info->section_size_bits);
> > >
> > > return TRUE;
> > > @@ -444,10 +486,33 @@ get_versiondep_info_arm64(void)
> > > return FALSE;
> > > }
> > >
> > > - info->page_offset = (0xffffffffffffffffUL) << (va_bits - 1);
> > > + /*
> > > + * See TCR_EL1, Translation Control Register (EL1) register
> > > + * description in the ARMv8 Architecture Reference Manual.
> > > + * Basically, we can use the TCR_EL1.T1SZ
> > > + * value to determine the virtual addressing range supported
> > > + * in the kernel-space (i.e. vabits_actual) since Linux 5.9.
> > > + */
> > > + if (NUMBER(TCR_EL1_T1SZ) != NOT_FOUND_NUMBER) {
> > > + vabits_actual = 64 - NUMBER(TCR_EL1_T1SZ);
> > > + DEBUG_MSG("vabits_actual : %d (vmcoreinfo)\n", vabits_actual);
> > > + } else if ((va_bits == 52) && (SYMBOL(mem_section) != NOT_FOUND_SYMBOL)) {
> > > + /*
> > > + * Linux 5.4 through 5.10 have the following linear space:
> > > + * 48-bit: 0xffff000000000000 - 0xffff7fffffffffff
> > > + * 58-bit: 0xfff0000000000000 - 0xfff7ffffffffffff
> > > + */
> > > + if (SYMBOL(mem_section) & (1UL << (52 - 1)))
> >
> > Sorry but I do not think any SYMBOL(x) is inside the range of linear
> > mapping address. All of them should be beyond kimage_vaddr.
As for this, SYMBOL(mem_section) is a bit special and shows the value in the
mem_section variable, not the address of it, by kernel commit a0b1280368d1.
Actually on my tiny arm64 board..
# strings /proc/kcore | less
...
OSRELEASE=5.4.0-1025-raspi
PAGESIZE=4096
SYMBOL(init_uts_ns)=ffffd6db0fa10568
SYMBOL(node_online_map)=ffffd6db0fa0a808
SYMBOL(swapper_pg_dir)=ffffd6db0f42b000
SYMBOL(_stext)=ffffd6db0e481000
SYMBOL(vmap_area_list)=ffffd6db0fa51680
SYMBOL(mem_map)=ffffd6db0fbc0240
SYMBOL(contig_page_data)=ffffd6db0faed6c0
SYMBOL(mem_section)=ffff0000eb809300 <<--
LENGTH(mem_section)=1024
SIZE(mem_section)=16
OFFSET(mem_section.section_mem_map)=0
...
NUMBER(VA_BITS)=48
So at least if the kernel is configured with CONFIG_SPARSEMEM_EXTREME,
I thought this would work. This is just a user tool effort to broaden
supported kernels, instead of necessary information from the kernel,
no need to support all kernels.
And the current makedumpfile's implementation, SYMBOL(mem_section) is set
to the address of the mem_section variable with -x option, so I wrote on
the cover letter that this patchset doesn't support "with -x option if
vabits_actual=52" case yet.
> >
> > Having vabits_actual is introduced and precise to resolve
> > is_lm_address(), but if it is not available, could we survive?
> >
> > _PAGE_OFFSET(52) < _PAGE_END(52) < _PAGE_OFFSET(48) < _PAGE_END(48)
> > Translating into numbers:
> > 0xfff0000000000000 < 0xfff8000000000000 < 0xffff000000000000 <
> > 0xffff800000000000
> >
> > Refer to linux/Documentation/arm64/memory.rst, the line
> > ffffa00010000000 fffff81ffffeffff ~88TB vmalloc
> >
> > It comes to the conclusion that any symbol > SYMBOL(_text) > _PAGE_END(48).
> >
> > So is_lm_address() can looks like
> > if (addr > _PAGE_END(48)), it is kimage
> > else, it is linear mapping
This sounds good and will work.
> >
> > So even more aggressive, we can exclude the need of vabits_actual
> > totally in this patch.
Interesting, but if vabits_actual is not available, I wonder how we can
get info->page_offset. Seems it's used mainly for p2v conversion and
is_kvaddr(), which still has a problem though.
>
> As Pingfan noted, this approach still has issues and it failed on my
> arm64 board in the limited testing I did today.
Can I have what configuration, error and debug messages you see,
for reference?
> I will do some more testing tomorrow and come back with an approach
> which addresses the same and also takes into account Pingfan's
> concerns.
ok, thanks!
Kazu
More information about the kexec
mailing list