Re: [PATCH v2] memblock: Add flags and nid info in memblock debugfs
From: Mike Rapoport
Date: Wed May 17 2023 - 09:42:33 EST
On Wed, May 17, 2023 at 11:37:25AM +0530, Anshuman Khandual wrote:
>
> On 5/17/23 08:27, Yuwei Guan wrote:
> > Currently, the memblock debugfs can display the count of memblock_type and
> > the base and end of the reg. However, when the following scenario occurs,
>
> scenarios where the memblock flags or nid varies inside a single PA range ?
> I guess the commit message description here can be improved to accommodate
> such details.
>
> > the information in the existing debugfs cannot make it clear why the
> > address is not consecutive.
> >
> > For example,
> > cat /sys/kernel/debug/memblock/memory
> > 0: 0x0000000080000000..0x00000000901fffff
> > 1: 0x0000000090200000..0x00000000905fffff
> > 2: 0x0000000090600000..0x0000000092ffffff
> > 3: 0x0000000093000000..0x00000000973fffff
> > 4: 0x0000000097400000..0x00000000b71fffff
> > 5: 0x00000000c0000000..0x00000000dfffffff
> > 6: 0x00000000e2500000..0x00000000f87fffff
> > 7: 0x00000000f8800000..0x00000000fa7fffff
> > 8: 0x00000000fa800000..0x00000000fd3effff
> > 9: 0x00000000fd3f0000..0x00000000fd3fefff
> > 10: 0x00000000fd3ff000..0x00000000fd7fffff
> > 11: 0x00000000fd800000..0x00000000fd901fff
> > 12: 0x00000000fd902000..0x00000000fd909fff
> > 13: 0x00000000fd90a000..0x00000000fd90bfff
> > 14: 0x00000000fd90c000..0x00000000ffffffff
> > 15: 0x0000000880000000..0x0000000affffffff
> >
> > So we can add flags and nid to this debugfs.
> >
> > For example,
> > cat /sys/kernel/debug/memblock/memory
> > cnt base..end flags nid
>
> These markers ^^^ are not aligned properly, and also might not be
> required as well.
>
> > 0: 0x0000000080000000..0x00000000901fffff 0x0 0x0
> > 1: 0x0000000090200000..0x00000000905fffff 0x4 0x0
> > 2: 0x0000000090600000..0x0000000092ffffff 0x0 0x0
> > 3: 0x0000000093000000..0x00000000973fffff 0x4 0x0
> > 4: 0x0000000097400000..0x00000000b71fffff 0x0 0x0
> > 5: 0x00000000c0000000..0x00000000dfffffff 0x0 0x0
> > 6: 0x00000000e2500000..0x00000000f87fffff 0x0 0x0
> > 7: 0x00000000f8800000..0x00000000fa7fffff 0x4 0x0
> > 8: 0x00000000fa800000..0x00000000fd3effff 0x0 0x0
> > 9: 0x00000000fd3f0000..0x00000000fd3fefff 0x4 0x0
> > 10: 0x00000000fd3ff000..0x00000000fd7fffff 0x0 0x0
> > 11: 0x00000000fd800000..0x00000000fd901fff 0x4 0x0
> > 12: 0x00000000fd902000..0x00000000fd909fff 0x0 0x0
> > 13: 0x00000000fd90a000..0x00000000fd90bfff 0x4 0x0
> > 14: 0x00000000fd90c000..0x00000000ffffffff 0x0 0x0
> > 15: 0x0000000880000000..0x0000000affffffff 0x0 0x0
> >
> > Signed-off-by: Yuwei Guan <ssawgyw@xxxxxxxxx>
> > Reviewed-by: Tarun Sahu <tsahu@xxxxxxxxxxxxx>
> > ---
> > mm/memblock.c | 8 ++++++--
> > 1 file changed, 6 insertions(+), 2 deletions(-)
> >
> > diff --git a/mm/memblock.c b/mm/memblock.c
> > index 511d4783dcf1..b36fb6b31e0f 100644
> > --- a/mm/memblock.c
> > +++ b/mm/memblock.c
> > @@ -2144,12 +2144,16 @@ static int memblock_debug_show(struct seq_file *m, void *private)
> > int i;
> > phys_addr_t end;
> >
> > + seq_puts(m, "cnt\tbase..end\tflags\tnid\n");
>
> Please drop this.
>
> > +
> > for (i = 0; i < type->cnt; i++) {
> > reg = &type->regions[i];
> > end = reg->base + reg->size - 1;
> >
> > - seq_printf(m, "%4d: ", i);
> > - seq_printf(m, "%pa..%pa\n", ®->base, &end);
> > + seq_printf(m, "%d:\t", i);
>
> Why drop the existing %4d formatting qualifier ?
>
> > + seq_printf(m, "%pa..%pa\t", ®->base, &end);
> > + seq_printf(m, "0x%x\t", reg->flags);
>
> Should there be ORed string values for each memblock flag ?
>
> enum memblock_flags {
> MEMBLOCK_NONE = 0x0, /* No special request */
> MEMBLOCK_HOTPLUG = 0x1, /* hotpluggable region */
> MEMBLOCK_MIRROR = 0x2, /* mirrored region */
> MEMBLOCK_NOMAP = 0x4, /* don't add to kernel direct mapping */
> MEMBLOCK_DRIVER_MANAGED = 0x8, /* always detected via a driver */
> };
>
> Something like NN | HT | MR | NM | DM ?
These are not less cryptic than numbers :)
Most of them are mutually exclusive, so maybe just spell them out fully,
just shorten DRIVER_MANAGED to DRV_MNG?
And make the flags dump the last to keep columns nicely aligned.
> > + seq_printf(m, "0x%x\n", memblock_get_region_node(reg));
> > }
> > return 0;
> > }
--
Sincerely yours,
Mike.