RE: [PATCH v2 2/3] hv: vmbus_post_msg: retry the hypercall on some transient errors

[Dexuan Cui]

> -----Original Message-----
> From: Jason Wang [mailto:jasowang@xxxxxxxxxx]
> Sent: Monday, February 2, 2015 17:41 PM
> To: Dexuan Cui
> Cc: gregkh@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; driverdev-
> devel@xxxxxxxxxxxxxxxxxxxxxx; olaf@xxxxxxxxx; apw@xxxxxxxxxxxxx; KY
> Srinivasan; vkuznets@xxxxxxxxxx; Haiyang Zhang
> Subject: Re: [PATCH v2 2/3] hv: vmbus_post_msg: retry the hypercall on some
> transient errors
> 
> On Mon, Feb 2, 2015 at 12:36 PM, Dexuan Cui <decui@xxxxxxxxxxxxx> wrote:
> > I got HV_STATUS_INVALID_CONNECTION_ID on Hyper-V 2008 R2 when
> keeping
> > running
> > "rmmod hv_netvsc; modprobe hv_netvsc; rmmod hv_utils; modprobe
> > hv_utils"
> > in a Linux guest. Looks the host has some kind of throttling
> > mechanism if
> > some kinds of hypercalls are sent too frequently.
> 
> Better to clarify this kind of throttling. Looks like it only affects
> HVCALL_POST_MESSAGE. Other looks good.
I'm not sure about the details either.
And it may be pretty host-specific, e.g., I can only get the error code on 
Hyper-V 2008 R2, but can't get it on 2012 R2. 

> Reviewed-by: Jason Wang <jasowang@xxxxxxxxxx>
Thanks!

> >
> > Without the patch, the driver can occasionally fail to load.
> >
> > Also let's retry HV_STATUS_INSUFFICIENT_MEMORY, though we didn't get
> > it
> > before.
> >
> > Removed 'case -ENOMEM', since the hypervisor doesn't return this.
> >
> > CC: "K. Y. Srinivasan" <kys@xxxxxxxxxxxxx>
> > Signed-off-by: Dexuan Cui <decui@xxxxxxxxxxxxx>
> > ---
> >
> > v2:
> >   updated the subject and changelog
> >   on HV_STATUS_INVALID_CONNECTION_ID: ret =  -EAGAIN;
> >   added HV_STATUS_INSUFFICIENT_MEMORY
> >   removed unreachable -ENOMEM
> >   changed the delay from 100ms to 1000ms
> >
> >  arch/x86/include/uapi/asm/hyperv.h |  2 ++
> >  drivers/hv/connection.c            | 11 +++++++++--
> >  2 files changed, 11 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/x86/include/uapi/asm/hyperv.h
> > b/arch/x86/include/uapi/asm/hyperv.h
> > index 90c458e..ce6068d 100644
> > --- a/arch/x86/include/uapi/asm/hyperv.h
> > +++ b/arch/x86/include/uapi/asm/hyperv.h
> > @@ -225,6 +225,8 @@
> >  #define HV_STATUS_INVALID_HYPERCALL_CODE	2
> >  #define HV_STATUS_INVALID_HYPERCALL_INPUT	3
> >  #define HV_STATUS_INVALID_ALIGNMENT		4
> > +#define HV_STATUS_INSUFFICIENT_MEMORY		11
> > +#define HV_STATUS_INVALID_CONNECTION_ID		18
> >  #define HV_STATUS_INSUFFICIENT_BUFFERS		19
> >
> >  typedef struct _HV_REFERENCE_TSC_PAGE {
> > diff --git a/drivers/hv/connection.c b/drivers/hv/connection.c
> > index c4acd1c..af2388f 100644
> > --- a/drivers/hv/connection.c
> > +++ b/drivers/hv/connection.c
> > @@ -440,9 +440,16 @@ int vmbus_post_msg(void *buffer, size_t buflen)
> >  		ret = hv_post_message(conn_id, 1, buffer, buflen);
> >
> >  		switch (ret) {
> > +		case HV_STATUS_INVALID_CONNECTION_ID:
> > +			/*
> > +			 * We could get this if we send messages too
> > +			 * frequently.
> > +			 */
> > +			ret = -EAGAIN;
> > +			break;
> > +		case HV_STATUS_INSUFFICIENT_MEMORY:
> >  		case HV_STATUS_INSUFFICIENT_BUFFERS:
> >  			ret = -ENOMEM;
> > -		case -ENOMEM:
> >  			break;
> >  		case HV_STATUS_SUCCESS:
> >  			return ret;
> > @@ -452,7 +459,7 @@ int vmbus_post_msg(void *buffer, size_t buflen)
> >  		}
> >
> >  		retries++;
> > -		msleep(100);
> > +		msleep(1000);
> >  	}
> >  	return ret;
> >  }
> > --
> > 1.9.1
> >