[RFC] Proportional bandwidth scheduling using anticipatory I/O scheduler on 2.6.24

2008-01-29 Thread Naveen Gupta
limits for various channels? What other things are expected? Would assigning priorities be of any use, either absolute priorities or soft priorities along with b/w limitations. I can add cgroup interfaces in another patch. Signed-off-by: Naveen Gupta <[EMAIL PROTECTED]> Index: l

Re: [PATCH] cgroup: limit block I/O bandwidth

2008-01-22 Thread Naveen Gupta
On 22/01/2008, Andrea Righi <[EMAIL PROTECTED]> wrote: > Naveen Gupta wrote: > > See if using priority levels to have per level bandwidth limit can > > solve the priority inversion problem you were seeing earlier. I have a > > priority scheduling patch for anticipator

Re: [PATCH] cgroup: limit block I/O bandwidth

2008-01-22 Thread Naveen Gupta
On 20/01/2008, Andrea Righi <[EMAIL PROTECTED]> wrote: > Jens Axboe wrote: > > On Sun, Jan 20 2008, Andrea Righi wrote: > >> Jens Axboe wrote: > >>> Your approach is totally flawed, imho. For instance, you don't want a > >>> process to be able to dirty memory at foo mb/sec but only actually > >>> w

Re: [PATCH] cgroup: limit block I/O bandwidth

2008-01-18 Thread Naveen Gupta
>Paul Menage wrote: >> On Jan 18, 2008 7:36 AM, Dhaval Giani <[EMAIL PROTECTED]> wrote: >>> On Fri, Jan 18, 2008 at 12:41:03PM +0100, Andrea Righi wrote: Allow to limit the block I/O bandwidth for specific process containers (cgroups) imposing additional delays on I/O requests for t

Re: [-mm PATCH] set enable bit instead of lock bit of Watchdog Timer in Intel 6300 chipset

2005-08-15 Thread Naveen Gupta
David, Yes, I have tested these patches. In fact I found these bugs while trying to make the driver work on our machines. -Naveen On Tue, 16 Aug 2005, David Härdeman wrote: > On Mon, Aug 15, 2005 at 02:21:05PM -0700, Naveen Gupta wrote: > > > >This patch sets the WDT_ENABLE

[-mm PATCH] remove use of pci_find_device in watchdog driver for Intel 6300ESB chipset

2005-08-15 Thread Naveen Gupta
This patch replaces obsolete 'pci_find_device' with 'pci_get_device' to prevent the device from being stolen under us in Watchdog timer driver for intel 6300ESB chipset. Signed-off-by: Naveen Gupta <[EMAIL PROTECTED]> Index: linux-2.6.12/driver

[-mm PATCH] set enable bit instead of lock bit of Watchdog Timer in Intel 6300 chipset

2005-08-15 Thread Naveen Gupta
This patch sets the WDT_ENABLE bit of the Lock Register to enable the watchdog and WDT_LOCK bit only if nowayout is set. The old code always sets the WDT_LOCK bit of watchdog timer for Intel 6300ESB chipset. So, we end up locking the watchdog instead of enabling it. Signed-off-by: Naveen Gupta

[-mm PATCH] set correct bit in reload register of Watchdog Timer for Intel 6300 chipset

2005-08-15 Thread Naveen Gupta
This patch writes into bit 8 of the reload register to perform the correct 'Reload Sequence' instead of writing into bit 4 of Watchdog for Intel 6300ESB chipset. Signed-off-by: Naveen Gupta <[EMAIL PROTECTED]> Index: linux-2.6.12/drivers/char/wat