On Tue, Nov 17, 2020 at 10:05:18AM +0200, Mike Rapoport wrote:
> On Mon, Nov 16, 2020 at 11:05:54PM +0300, Serge Semin wrote:
> > On Mon, Nov 16, 2020 at 06:45:15PM +0100, Thomas Bogendoerfer wrote:
> > > The loop over all memblocks works with PFN numbers and not physical
> > > addresses, so we need for_each_mem_pfn_range().
> >
> > Great catch! Don't know how that has been working so far. Anyway
>
> The loop is relevant only for systems with highmem, apparently there are
> not many highmem users out there.
That has been mostly a rhetorical question. The thing is that our
platform is 32bit and it is active user of highmem.) So I am very
puzzled how I haven't noticed a effect caused by that bug before.
Most likely that happened due to our platform having a first memory
chunk starting with physical address 0x0. That's why we always have a
low memory defined.
-Sergey
>
> > Reviewed-by: Serge Semin <fancer.lancer@xxxxxxxxx>
> >
> > -Sergey
> >
> > >
> > > Fixes: b10d6bca8720 ("arch, drivers: replace for_each_membock() with for_each_mem_range()")
> > > Signed-off-by: Thomas Bogendoerfer <tsbogend@xxxxxxxxxxxxxxxx>
> > > ---
> > > arch/mips/kernel/setup.c | 6 +++---
> > > 1 file changed, 3 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/arch/mips/kernel/setup.c b/arch/mips/kernel/setup.c
> > > index 0d4253208bde..ca579deef939 100644
> > > --- a/arch/mips/kernel/setup.c
> > > +++ b/arch/mips/kernel/setup.c
> > > @@ -262,8 +262,8 @@ static void __init bootmem_init(void)
> > > static void __init bootmem_init(void)
> > > {
> > > phys_addr_t ramstart, ramend;
> > > - phys_addr_t start, end;
> > > - u64 i;
> > > + unsigned long start, end;
> > > + int i;
> > >
> > > ramstart = memblock_start_of_DRAM();
> > > ramend = memblock_end_of_DRAM();
> > > @@ -300,7 +300,7 @@ static void __init bootmem_init(void)
> > >
> > > min_low_pfn = ARCH_PFN_OFFSET;
> > > max_pfn = PFN_DOWN(ramend);
> > > - for_each_mem_range(i, &start, &end) {
> > > + for_each_mem_pfn_range(i, MAX_NUMNODES, &start, &end, NULL) {
> > > /*
> > > * Skip highmem here so we get an accurate max_low_pfn if low
> > > * memory stops short of high memory.
> > > --
> > > 2.16.4
> > >
>
> --
> Sincerely yours,
> Mike.
