Re: [PATCH v4 2/3] ACPI / NUMA: Add warning message if the padding size for KASLR is not enough

[Masayoshi Mizuma]

On Fri, Sep 28, 2018 at 10:48:57AM +0800, Baoquan He wrote:
> On 09/27/18 at 04:31pm, Masayoshi Mizuma wrote:
> > From: Masayoshi Mizuma <m.mizuma@xxxxxxxxxxxxxx>
> > 
> > Add warning message if the padding size for KASLR,
> > rand_mem_physical_padding, is not enough. The message also
> > says the suitable padding size.
> > 
> > Signed-off-by: Masayoshi Mizuma <m.mizuma@xxxxxxxxxxxxxx>
> > ---
> >  arch/x86/include/asm/setup.h |  2 ++
> >  drivers/acpi/numa.c          | 14 ++++++++++++++
> >  2 files changed, 16 insertions(+)
> > 
> > diff --git a/arch/x86/include/asm/setup.h b/arch/x86/include/asm/setup.h
> > index ae13bc9..65a5bf8 100644
> > --- a/arch/x86/include/asm/setup.h
> > +++ b/arch/x86/include/asm/setup.h
> > @@ -80,6 +80,8 @@ static inline unsigned long kaslr_offset(void)
> >  	return (unsigned long)&_text - __START_KERNEL;
> >  }
> >  
> > +extern int rand_mem_physical_padding;
> > +
> >  /*
> >   * Do NOT EVER look at the BIOS memory size location.
> >   * It does not work on many machines.
> > diff --git a/drivers/acpi/numa.c b/drivers/acpi/numa.c
> > index 8516760..9c3cc3c 100644
> > --- a/drivers/acpi/numa.c
> > +++ b/drivers/acpi/numa.c
> > @@ -32,6 +32,7 @@
> >  #include <linux/numa.h>
> >  #include <linux/nodemask.h>
> >  #include <linux/topology.h>
> > +#include <asm/setup.h>
> >  
> >  static nodemask_t nodes_found_map = NODE_MASK_NONE;
> >  
> > @@ -435,6 +436,8 @@ acpi_table_parse_srat(enum acpi_srat_type id,
> >  int __init acpi_numa_init(void)
> >  {
> >  	int cnt = 0;
> > +	u32 max_phys_addr_tb;
> > +	u64 max_phys_addr;
> >  
> >  	if (acpi_disabled)
> >  		return -EINVAL;
> > @@ -463,6 +466,17 @@ int __init acpi_numa_init(void)
> >  
> >  		cnt = acpi_table_parse_srat(ACPI_SRAT_TYPE_MEMORY_AFFINITY,
> >  					    acpi_parse_memory_affinity, 0);
> > +
> > +		if (parsed_numa_memblks && kaslr_enabled()) {
> > +			max_phys_addr = PFN_PHYS(max_possible_pfn);
> > +			max_phys_addr_tb = (roundup(max_phys_addr, 1ULL << 40)) >> 40;
> > +
> > +			if (max_phys_addr_tb > rand_mem_physical_padding)
> 
> Here I assume max_phys_addr_tb is the end of the possible RAM in system.

Yes, correct.

> rand_mem_physical_padding is the preserved space for later memory
> extending. Don't we add the actual RAM size to the
> rand_mem_physical_padding, then compare with max_phys_addr_tb?

This is very good point! You are right, the padding should be
adjusted to exceed the possible RAM size, like as follows.

   ADDRESS
     0
     :
    1TB           <= actual RAM size
     :
    2TB           <= possible RAM size
    2TB and more  <= actual + padding size (kaslr_regions[0].size_tb)

I'll fix it, thanks!

- Masa

> 
> Please correct me if I am wrong.
> 
> Thanks
> Baoquan
> 
> > +				pr_warn("Set 'rand_mem_physical_padding=%d' "
> > +					"as the kernel parameter. "
> > +					"Otherwise, memory hotadd may be failed.\n",
> > +					max_phys_addr_tb);
> > +		}
> >  	}
> >  
> >  	/* SLIT: System Locality Information Table */
> > -- 
> > 2.18.0
> >