Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Fri, Feb 12, 2016 at 02:49:38AM +0530, Parav Pandit wrote: > 1. Removed two type of resource pool, made is single type (as you > described in past comment) > 2. Removed match tokens and have array definition like "qp", "mr", "cq" etc. > 3. Wrote small parser and avoided

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Fri, Feb 12, 2016 at 02:49:38AM +0530, Parav Pandit wrote: > 1. Removed two type of resource pool, made is single type (as you > described in past comment) > 2. Removed match tokens and have array definition like "qp", "mr", "cq" etc. > 3. Wrote small parser and avoided

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hi Tejun, (Sending again as by mistake previous mail was non plain text, sorry about that). So based on your comments, Haggai's comments below and past experience, I will spin v6.o I have made changes, and testing them. I am likely to post during coming long weekend. I have simplified many

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On Wed, Feb 10, 2016 at 9:57 PM, Haggai Eran wrote: > > There is indeed an effort to backport the latest RDMA subsystem modules to > older kernels, and it would be preferable to be able to introduce new > resources through these modules. Right. There is hardly 10 to 20 lines of code that allows

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hi Tejun, (Sending again as by mistake previous mail was non plain text, sorry about that). So based on your comments, Haggai's comments below and past experience, I will spin v6.o I have made changes, and testing them. I am likely to post during coming long weekend. I have simplified many

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On Wed, Feb 10, 2016 at 9:57 PM, Haggai Eran wrote: > > There is indeed an effort to backport the latest RDMA subsystem modules to > older kernels, and it would be preferable to be able to introduce new > resources through these modules. Right. There is hardly 10 to 20 lines

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On 01/02/2016 20:59, Parav Pandit wrote: > On Tue, Feb 2, 2016 at 12:10 AM, Tejun Heo wrote: >> So, I'm really not gonna go for individual drivers defining resources >> on their own. That's a trainwreck waiting to happen. There needs to >> be a lot more scrutiny than that. >> > Not every low

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On 01/02/2016 20:59, Parav Pandit wrote: > On Tue, Feb 2, 2016 at 12:10 AM, Tejun Heo wrote: >> So, I'm really not gonna go for individual drivers defining resources >> on their own. That's a trainwreck waiting to happen. There needs to >> be a lot more scrutiny than that. >> >

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On Tue, Feb 2, 2016 at 12:10 AM, Tejun Heo wrote: > Hello, Parav. > > On Sun, Jan 31, 2016 at 11:20:45PM +0530, Parav Pandit wrote: >> > Yes, it can. It just becomes a part of kernel ABI that the IB stack >> > module depends on. >> > >> o.k. Its doable, but I believe its simple but its

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Sun, Jan 31, 2016 at 11:20:45PM +0530, Parav Pandit wrote: > > Yes, it can. It just becomes a part of kernel ABI that the IB stack > > module depends on. > > > o.k. Its doable, but I believe its simple but its restrictive. The whole point is to make it somewhat restrictive so

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On Tue, Feb 2, 2016 at 12:10 AM, Tejun Heo wrote: > Hello, Parav. > > On Sun, Jan 31, 2016 at 11:20:45PM +0530, Parav Pandit wrote: >> > Yes, it can. It just becomes a part of kernel ABI that the IB stack >> > module depends on. >> > >> o.k. Its doable, but I believe its simple

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Sun, Jan 31, 2016 at 11:20:45PM +0530, Parav Pandit wrote: > > Yes, it can. It just becomes a part of kernel ABI that the IB stack > > module depends on. > > > o.k. Its doable, but I believe its simple but its restrictive. The whole point is to make it somewhat restrictive so

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hi Doug, Liran, Jason, How would you like to see RDMA verb resources being defined - in RDMA cgroup or in IB stack? In current patch v5, its defined by the IB stack which is often shipped as different package due to high amount of changes, bug fixes, features. In v0 patch it was defined by the

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Sun, Jan 31, 2016 at 04:11:54PM +0530, Parav Pandit wrote: > No. We agreed that let IB stack define in the header file that rdmacg > can include. > However when I started with that I realized that, such design has > basic flaw that IB stack is compiled as loadable modules. >

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On Sun, Jan 31, 2016 at 3:32 PM, Tejun Heo wrote: > Hello, Parav. > > On Sun, Jan 31, 2016 at 02:14:13AM +0530, Parav Pandit wrote: > ... >> V1 patch had IB resources defined in the header file of rdmacg, which >> I believe is very restrictive model with evolving rdma stack and >> features. > >

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Sun, Jan 31, 2016 at 02:14:13AM +0530, Parav Pandit wrote: ... > V1 patch had IB resources defined in the header file of rdmacg, which > I believe is very restrictive model with evolving rdma stack and > features. Wasn't this the model that we agreed upon? Besides, even if the

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hi Doug, Liran, Jason, How would you like to see RDMA verb resources being defined - in RDMA cgroup or in IB stack? In current patch v5, its defined by the IB stack which is often shipped as different package due to high amount of changes, bug fixes, features. In v0 patch it was defined by the

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

On Sun, Jan 31, 2016 at 3:32 PM, Tejun Heo wrote: > Hello, Parav. > > On Sun, Jan 31, 2016 at 02:14:13AM +0530, Parav Pandit wrote: > ... >> V1 patch had IB resources defined in the header file of rdmacg, which >> I believe is very restrictive model with evolving rdma stack and

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Sun, Jan 31, 2016 at 02:14:13AM +0530, Parav Pandit wrote: ... > V1 patch had IB resources defined in the header file of rdmacg, which > I believe is very restrictive model with evolving rdma stack and > features. Wasn't this the model that we agreed upon? Besides, even if the

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, Parav. On Sun, Jan 31, 2016 at 04:11:54PM +0530, Parav Pandit wrote: > No. We agreed that let IB stack define in the header file that rdmacg > can include. > However when I started with that I realized that, such design has > basic flaw that IB stack is compiled as loadable modules. >

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hi Tejun, On Sun, Jan 31, 2016 at 12:00 AM, Tejun Heo wrote: > Hello, > > On Sat, Jan 30, 2016 at 08:53:05PM +0530, Parav Pandit wrote: >> +#ifdef CONFIG_CGROUP_RDMA >> +#define RDMACG_MAX_RESOURCE_INDEX (64) > > The list of resources are known at compile time. Why is this separate > fixed

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, On Sat, Jan 30, 2016 at 08:53:05PM +0530, Parav Pandit wrote: > +#ifdef CONFIG_CGROUP_RDMA > +#define RDMACG_MAX_RESOURCE_INDEX (64) The list of resources are known at compile time. Why is this separate fixed number necessary? > +struct match_token; There's no circular dependency issue

[PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Added rdma cgroup controller that does accounting, limit enforcement on rdma/IB verbs and hw resources. Added rdma cgroup header file which defines its APIs to perform charing/uncharing functionality and device registration which will participate in controller functions of accounting and limit

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hi Tejun, On Sun, Jan 31, 2016 at 12:00 AM, Tejun Heo wrote: > Hello, > > On Sat, Jan 30, 2016 at 08:53:05PM +0530, Parav Pandit wrote: >> +#ifdef CONFIG_CGROUP_RDMA >> +#define RDMACG_MAX_RESOURCE_INDEX (64) > > The list of resources are known at compile time. Why is this

Re: [PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Hello, On Sat, Jan 30, 2016 at 08:53:05PM +0530, Parav Pandit wrote: > +#ifdef CONFIG_CGROUP_RDMA > +#define RDMACG_MAX_RESOURCE_INDEX (64) The list of resources are known at compile time. Why is this separate fixed number necessary? > +struct match_token; There's no circular dependency issue

[PATCHv3 1/3] rdmacg: Added rdma cgroup controller.

Added rdma cgroup controller that does accounting, limit enforcement on rdma/IB verbs and hw resources. Added rdma cgroup header file which defines its APIs to perform charing/uncharing functionality and device registration which will participate in controller functions of accounting and limit