Re: [PATCH] powerpc: align memory_limit to 16MB in early_parse_mem

From: Joel Savitz
Date: Mon Apr 01 2024 - 10:17:52 EST


On Tue, Mar 26, 2024 at 12:45 AM Joel Savitz <jsavitz@xxxxxxxxxx> wrote:
>
> On Fri, Mar 8, 2024 at 5:18 AM Aneesh Kumar K.V <aneesh.kumar@xxxxxxxxxx> wrote:
> >
> > Joel Savitz <jsavitz@xxxxxxxxxx> writes:
> >
> > > On 64-bit powerpc, usage of a non-16MB-aligned value for the mem= kernel
> > > cmdline parameter results in a system hang at boot.
> > >
> > > For example, using 'mem=4198400K' will always reproduce this issue.
> > >
> > > This patch fixes the problem by aligning any argument to mem= to 16MB
> > > corresponding with the large page size on powerpc.
> > >
> > > Fixes: 2babf5c2ec2f ("[PATCH] powerpc: Unify mem= handling")
> > > Co-developed-by: Gonzalo Siero <gsierohu@xxxxxxxxxx>
> > > Signed-off-by: Gonzalo Siero <gsierohu@xxxxxxxxxx>
> > > Signed-off-by: Joel Savitz <jsavitz@xxxxxxxxxx>
> > > ---
> > > arch/powerpc/kernel/prom.c | 6 +++++-
> > > 1 file changed, 5 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
> > > index 0b5878c3125b..8cd3e2445d8a 100644
> > > --- a/arch/powerpc/kernel/prom.c
> > > +++ b/arch/powerpc/kernel/prom.c
> > > @@ -82,8 +82,12 @@ static int __init early_parse_mem(char *p)
> > > {
> > > if (!p)
> > > return 1;
> > > -
> > > +#ifdef CONFIG_PPC64
> > > + /* Align to 16 MB == size of ppc64 large page */
> > > + memory_limit = ALIGN(memparse(p, &p), 0x1000000);
> > > +#else
> > > memory_limit = PAGE_ALIGN(memparse(p, &p));
> > > +#endif
> > > DBG("memory limit = 0x%llx\n", memory_limit);
> > >
> > > return 0;
> > > --
> > > 2.43.0
> >
> > Can you try this change?
> >
> > commit 5555bc55e1aa71f545cff31e1eccdb4a2e39df84
> > Author: Aneesh Kumar K.V (IBM) <aneesh.kumar@xxxxxxxxxx>
> > Date: Fri Mar 8 14:45:26 2024 +0530
> >
> > powerpc/mm: Align memory_limit value specified using mem= kernel parameter
> >
> > The value specified for the memory limit is used to set a restriction on
> > memory usage. It is important to ensure that this restriction is within
> > the linear map kernel address space range. The hash page table
> > translation uses a 16MB page size to map the kernel linear map address
> > space. htab_bolt_mapping() function aligns down the size of the range
> > while mapping kernel linear address space. Since the memblock limit is
> > enforced very early during boot, before we can detect the type of memory
> > translation (radix vs hash), we align the memory limit value specified
> > as a kernel parameter to 16MB. This alignment value will work for both
> > hash and radix translations.
> >
> > Signed-off-by: Aneesh Kumar K.V (IBM) <aneesh.kumar@xxxxxxxxxx>
> >
> > diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
> > index 0b5878c3125b..9bd965d35352 100644
> > --- a/arch/powerpc/kernel/prom.c
> > +++ b/arch/powerpc/kernel/prom.c
> > @@ -824,8 +824,11 @@ void __init early_init_devtree(void *params)
> > reserve_crashkernel();
> > early_reserve_mem();
> >
> > - /* Ensure that total memory size is page-aligned. */
> > - limit = ALIGN(memory_limit ?: memblock_phys_mem_size(), PAGE_SIZE);
> > + if (memory_limit > memblock_phys_mem_size())
> > + memory_limit = 0;
> > +
> > + /* Align down to 16 MB which is large page size with hash page translation */
> > + limit = ALIGN_DOWN(memory_limit ?: memblock_phys_mem_size(), SZ_16M);
> > memblock_enforce_memory_limit(limit);
> >
> > #if defined(CONFIG_PPC_BOOK3S_64) && defined(CONFIG_PPC_4K_PAGES)
> > diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c
> > index e67effdba85c..d6410549e141 100644
> > --- a/arch/powerpc/kernel/prom_init.c
> > +++ b/arch/powerpc/kernel/prom_init.c
> > @@ -817,8 +817,8 @@ static void __init early_cmdline_parse(void)
> > opt += 4;
> > prom_memory_limit = prom_memparse(opt, (const char **)&opt);
> > #ifdef CONFIG_PPC64
> > - /* Align to 16 MB == size of ppc64 large page */
> > - prom_memory_limit = ALIGN(prom_memory_limit, 0x1000000);
> > + /* Align down to 16 MB which is large page size with hash page translation */
> > + prom_memory_limit = ALIGN_DOWN(prom_memory_limit, SZ_16M);
> > #endif
> > }
> >
> >
>
> Sorry for the delayed reply. I just tested this patch and it fixes the
> bug for me.

Hi,

Just a quick follow up on this.

The above patch fixed the bug for me.

How do we want to proceed?

Best,
Joel Savitz