Re: Enabling peer to peer device transactions for PCIe devices

On Fri, Jan 06, 2017 at 11:56:30AM -0500, Serguei Sagalovitch wrote: > On 2017-01-05 08:58 PM, Jerome Glisse wrote: > > On Thu, Jan 05, 2017 at 05:30:34PM -0700, Jason Gunthorpe wrote: > > > On Thu, Jan 05, 2017 at 06:23:52PM -0500, Jerome Glisse wrote: > > > > > > > > I still don't understand

Re: Enabling peer to peer device transactions for PCIe devices

On Fri, Jan 06, 2017 at 12:37:22PM -0500, Jerome Glisse wrote: > On Fri, Jan 06, 2017 at 11:56:30AM -0500, Serguei Sagalovitch wrote: > > On 2017-01-05 08:58 PM, Jerome Glisse wrote: > > > On Thu, Jan 05, 2017 at 05:30:34PM -0700, Jason Gunthorpe wrote: > > > > On Thu, Jan 05, 2017 at 06:23:52PM

[PATCH v4] libnvdimm: clear poison in mem map metadata

Clearing out the poison in the metadata block of the namespace before we use it. Range from start + 8k to pfn_sb->dataoff. Signed-off-by: Dave Jiang --- drivers/nvdimm/pfn_devs.c | 25 + 1 file changed, 25 insertions(+) diff --git

Re: [PATCH v2 0/4] Write protect DAX PMDs in *sync path

On Thu, Jan 05, 2017 at 05:27:34PM -0800, Andrew Morton wrote: > On Tue, 3 Jan 2017 17:13:49 -0700 Ross Zwisler > wrote: > > > On Thu, Dec 22, 2016 at 02:18:52PM -0700, Ross Zwisler wrote: > > > Currently dax_mapping_entry_mkclean() fails to clean and write protect

RE: Enabling peer to peer device transactions for PCIe devices

> -Original Message- > From: Jason Gunthorpe [mailto:jguntho...@obsidianresearch.com] > Sent: Friday, January 06, 2017 1:26 PM > To: Jerome Glisse > Cc: Sagalovitch, Serguei; Jerome Glisse; Deucher, Alexander; 'linux- > ker...@vger.kernel.org'; 'linux-r...@vger.kernel.org'; 'linux- >

Re: Enabling peer to peer device transactions for PCIe devices

On 06/01/17 11:26 AM, Jason Gunthorpe wrote: > Make a generic API for all of this and you'd have my vote.. > > IMHO, you must support basic pinning semantics - that is necessary to > support generic short lived DMA (eg filesystem, etc). That hardware > can clearly do that if it can support

Re: [PATCH v5] x86: fix kaslr and memmap collision

On Thu, Jan 5, 2017 at 6:44 PM, Baoquan He wrote: > Add Kees to let him have a look at this too. > > On 01/05/17 at 05:21pm, Baoquan He wrote: >> On 01/04/17 at 11:29am, Dave Jiang wrote: >> > CONFIG_RANDOMIZE_BASE relocates the kernel to a random base address. >> > However it

[RFC PATCH v2 2/2] block: fix blk_get_backing_dev_info() crash, use bdev->bd_queue

The ->bd_queue member of struct block_device was added in commit 87192a2a49c4 ("vfs: cache request_queue in struct block_device") in v3.3. However, blk_get_backing_dev_info() has been using bdev_get_queue() and grabbing the request_queue through the gendisk since before the git era. At final

[RFC PATCH v2 1/2] block: fix lifetime of request_queue / backing_dev_info relative to bdev

By definition the lifetime of a struct block_device is equal to the lifetime of its corresponding inode since they both live in struct bdev_inode. Up until the inode is destroyed it may be the target of delayed write-back requests. Issuing write-back to a block_device requires a lookup of the bdev

[ndctl PATCH 7/7] ndctl, list: limit mappings when --dimm is specified

Single out the mapping for the requested dimm device when specified. Signed-off-by: Dan Williams --- ndctl/builtin-list.c |3 +++ 1 file changed, 3 insertions(+) diff --git a/ndctl/builtin-list.c b/ndctl/builtin-list.c index ed4edf6f0249..e8d007003b62 100644 ---

[ndctl PATCH 0/7] introduce 'daxctl list', and 'ndctl list' updates

* The 'ndctl list' command awkwardly prints out all the corresponding device-dax information when a namespace is in 'dax' mode. Conversely if someone is only interested in listing device-dax information they need to contend with libnvdimm data. Introduce a separate daxctl utility with its

[ndctl PATCH 1/7] ndctl, daxctl: refactor main boilerplate for a new 'daxctl' utility

Prepare the generic command infrastructure for reuse with daxctl. Signed-off-by: Dan Williams --- Makefile.am |2 ndctl/Makefile.am |1 ndctl/builtin-bat.c |2 ndctl/builtin-create-nfit.c |2

[ndctl PATCH 6/7] ndctl, {enable,disable}-region: filter by type

Allow the region to be selected by type. For example, disable all pmem: ndctl disable-region -t pmem all One case this is useful for is scripting the bring-up of individual namespaces rather than the default of all at once and in parallel. For example: * blacklist the nd_pmem module *

[ndctl PATCH 3/7] ndctl, list: add option to filter namespace by mode

For example, "ndctl list -m dax" list all the namespaces that are hosting a device-dax region. Signed-off-by: Dan Williams --- ndctl/builtin-list.c | 33 - ndctl/libndctl.h.in |1 + 2 files changed, 33 insertions(+), 1 deletion(-)

[ndctl PATCH 4/7] ndctl, list: add '--device-dax' option

Only include dax region and sub-device info when requested by this new option to ndctl list. If '--device-dax' is the only option then the listing will only include namespaces in device-dax mode. Signed-off-by: Dan Williams --- ndctl/builtin-list.c |7

Re: Enabling peer to peer device transactions for PCIe devices

Hello, I've been watching this thread not as a kernel developer, but as an user interested in doing peer-to-peer access between network card and GPU. I believe that merging raw direct access with vma overcomplicates things for our use case. We'll have a very large camera streaming data at high