From: George Dunlap <george.dunlap@eu.citrix.com>
To: David Vrabel <david.vrabel@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>,
Ian Jackson <Ian.Jackson@eu.citrix.com>,
George Dunlap <george.dunlap@citrix.com>,
xen-devel@lists.xen.org
Subject: Re: [PATCH 3/3] libxc: add xc_tbuf_trace() to insert trace records
Date: Thu, 15 Aug 2013 15:37:10 +0100 [thread overview]
Message-ID: <520CE796.1090603@eu.citrix.com> (raw)
In-Reply-To: <1374758582-29038-4-git-send-email-david.vrabel@citrix.com>
On 25/07/13 14:23, David Vrabel wrote:
> From: David Vrabel <david.vrabel@citrix.com>
>
> Add xc_tbuf_trace() to allow trace records to be added to the trace
> buffer. The subclass and event number and up to 7 uin32_t arguments
> may be specified.
>
> The hypercall sub-op used is HVMOP_xentrace which (despite the name)
> may be used by PV guests.
Would it make sense to make this interface more like the hypervisor's
trace_var() -- that is, taking a sizeof() and single pointer, so that
callers can easily pass in packed structs?
I suppose there's no reason we couldn't add a function with that
interface at some point in the future if people want it. Any thoughts
from the toolstack maintainers?
-George
>
> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
> ---
> tools/libxc/xc_tbuf.c | 39 +++++++++++++++++++++++++++++++++++++++
> tools/libxc/xenctrl.h | 11 +++++++++++
> xen/arch/x86/hvm/hvm.c | 2 +-
> 3 files changed, 51 insertions(+), 1 deletions(-)
>
> diff --git a/tools/libxc/xc_tbuf.c b/tools/libxc/xc_tbuf.c
> index 4fb7bb1..0644952 100644
> --- a/tools/libxc/xc_tbuf.c
> +++ b/tools/libxc/xc_tbuf.c
> @@ -156,3 +156,42 @@ int xc_tbuf_set_evt_mask(xc_interface *xch, uint32_t mask)
> return do_sysctl(xch, &sysctl);
> }
>
> +int xc_tbuf_trace(xc_interface *xch, uint16_t event, unsigned nr_args, ...)
> +{
> + DECLARE_HYPERCALL;
> + DECLARE_HYPERCALL_BUFFER(xen_hvm_xentrace_t, trace);
> + unsigned i;
> + va_list args;
> + int ret = -1;
> +
> + if ( nr_args > 7 ) {
> + errno = -EINVAL;
> + goto out;
> + }
> +
> + trace = xc_hypercall_buffer_alloc(xch, trace, sizeof(*trace));
> + if ( trace == NULL )
> + {
> + PERROR("Count not alloc bounce buffer for trace hypercall");
> + goto out;
> + }
> +
> + trace->event = event;
> + trace->extra_bytes = nr_args * sizeof(uint32_t);
> +
> + va_start(args, nr_args);
> + for (i = 0; i < nr_args; i++)
> + ((uint32_t *)trace->extra)[i] = va_arg(args, uint32_t);
> + va_end(args);
> +
> + hypercall.op = __HYPERVISOR_hvm_op;
> + hypercall.arg[0] = HVMOP_xentrace;
> + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(trace);
> +
> + ret = do_xen_hypercall(xch, &hypercall);
> +
> +out:
> + xc_hypercall_buffer_free(xch, trace);
> +
> + return ret;
> +}
> diff --git a/tools/libxc/xenctrl.h b/tools/libxc/xenctrl.h
> index 388a9c3..927b198 100644
> --- a/tools/libxc/xenctrl.h
> +++ b/tools/libxc/xenctrl.h
> @@ -1387,6 +1387,17 @@ int xc_tbuf_set_cpu_mask(xc_interface *xch, uint32_t mask);
>
> int xc_tbuf_set_evt_mask(xc_interface *xch, uint32_t mask);
>
> +/**
> + * Insert a trace record into the trace buffer.
> + *
> + * The trace records use the TRC_GUEST class and 'event' sets the
> + * sub-class and the event number. There are no predefined values for
> + * these.
> + *
> + * Up to 7 uint32_t arguments may be included in the trace record.
> + */
> +int xc_tbuf_trace(xc_interface *xch, uint16_t event, unsigned nr_args, ...);
> +
> int xc_domctl(xc_interface *xch, struct xen_domctl *domctl);
> int xc_sysctl(xc_interface *xch, struct xen_sysctl *sysctl);
>
> diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> index b0d8094..b24a37d 100644
> --- a/xen/arch/x86/hvm/hvm.c
> +++ b/xen/arch/x86/hvm/hvm.c
> @@ -4388,7 +4388,7 @@ long do_hvm_op(unsigned long op, XEN_GUEST_HANDLE_PARAM(void) arg)
> return -EFAULT;
>
> if ( tr.extra_bytes > sizeof(tr.extra)
> - || (tr.event & ~((1u<<TRC_CLS_SHIFT)-1)) )
> + || (tr.event & ~((1u<<TRC_SUBCLS_SHIFT)-1)) )
> return -EINVAL;
>
> trace_var(TRC_GUEST_EVENT(tr.event), 1 /*cycles*/,
next prev parent reply other threads:[~2013-08-15 14:37 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-07-25 13:22 [PATCH 0/3] xen: guest tracing improvements David Vrabel
2013-07-25 13:23 ` [PATCH 1/3] trace: include timestamp in trace records added by HVMOP_xentrace David Vrabel
2013-08-15 14:15 ` George Dunlap
2013-07-25 13:23 ` [PATCH 2/3] trace: allow HVMOP_xentrace to set trace record subclass David Vrabel
2013-08-15 14:18 ` George Dunlap
2013-07-25 13:23 ` [PATCH 3/3] libxc: add xc_tbuf_trace() to insert trace records David Vrabel
2013-07-25 13:39 ` David Vrabel
2013-08-15 14:37 ` George Dunlap [this message]
2013-08-15 14:39 ` David Vrabel
2013-08-15 14:48 ` George Dunlap
2013-08-06 14:43 ` [PATCH 0/3] xen: guest tracing improvements David Vrabel
2013-08-06 14:46 ` George Dunlap
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=520CE796.1090603@eu.citrix.com \
--to=george.dunlap@eu.citrix.com \
--cc=Ian.Jackson@eu.citrix.com \
--cc=david.vrabel@citrix.com \
--cc=george.dunlap@citrix.com \
--cc=ian.campbell@citrix.com \
--cc=xen-devel@lists.xen.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).