linux-kernel - Re: [PATCH 3/3] libnvdimm, pmem: clear poison on write

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 11 Mar 2016 00:39:45 +0000
From:	"Verma, Vishal L" <vishal.l.verma@...el.com>
To:	"Williams, Dan J" <dan.j.williams@...el.com>,
	"linux-nvdimm@...ts.01.org" <linux-nvdimm@...ts.01.org>
CC:	"ross.zwisler@...ux.intel.com" <ross.zwisler@...ux.intel.com>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"x86@...nel.org" <x86@...nel.org>,
	"linux-acpi@...r.kernel.org" <linux-acpi@...r.kernel.org>
Subject: Re: [PATCH 3/3] libnvdimm, pmem: clear poison on write

On Tue, 2016-03-08 at 14:47 -0800, Dan Williams wrote:
> If a write is directed at a known bad block perform the following:
> 
> 1/ write the data
> 
> 2/ send a clear poison command
> 
> 3/ invalidate the poison out of the cache hierarchy
> 
> Cc: <x86@...nel.org>
> Cc: Vishal Verma <vishal.l.verma@...el.com>
> Cc: Ross Zwisler <ross.zwisler@...ux.intel.com>
> Signed-off-by: Dan Williams <dan.j.williams@...el.com>
> ---
>  arch/x86/include/asm/pmem.h |    5 +++++
>  drivers/nvdimm/bus.c        |   46
> +++++++++++++++++++++++++++++++++++++++++++
>  drivers/nvdimm/nd.h         |    2 ++
>  drivers/nvdimm/pmem.c       |   29 ++++++++++++++++++++++++++-
>  include/linux/pmem.h        |   19 ++++++++++++++++++
>  5 files changed, 100 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/include/asm/pmem.h b/arch/x86/include/asm/pmem.h
> index c57fd1ea9689..bf8b35d2035a 100644
> --- a/arch/x86/include/asm/pmem.h
> +++ b/arch/x86/include/asm/pmem.h

<>

>  static int pmem_do_bvec(struct pmem_device *pmem, struct page *page,
>  			unsigned int len, unsigned int off, int rw,
>  			sector_t sector)
>  {
>  	int rc = 0;
> +	bool bad_pmem = false;
>  	void *mem = kmap_atomic(page);
>  	phys_addr_t pmem_off = sector * 512 + pmem->data_offset;
>  	void __pmem *pmem_addr = pmem->virt_addr + pmem_off;
>  
> +	if (unlikely(is_bad_pmem(&pmem->bb, sector, len)))
> +		bad_pmem = true;
> +
>  	if (rw == READ) {
> -		if (unlikely(is_bad_pmem(&pmem->bb, sector, len)))
> +		if (unlikely(bad_pmem))
>  			rc = -EIO;
>  		else {
>  			memcpy_from_pmem(mem + off, pmem_addr, len);
> @@ -81,6 +104,10 @@ static int pmem_do_bvec(struct pmem_device *pmem,
> struct page *page,
>  	} else {
>  		flush_dcache_page(page);
>  		memcpy_to_pmem(pmem_addr, mem + off, len);
> +		if (unlikely(bad_pmem)) {
> +			pmem_clear_poison(pmem, pmem_off, len);
> +			memcpy_to_pmem(pmem_addr, mem + off, len);
> +		}
>  	}

Just noticed this -- why do we memcpy_to_pmem twice in the error case?
Sh
ouldn't it be:

	if (unlikely(bad_pmem))
		pmem_clear_poison(pmem, pmem_off, len);
	memcpy_to_pmem(pmem_addr, mem + off, len);