From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Vrabel Subject: Re: [PATCH 1/2] trace: improve usefulness of hypercall trace record Date: Mon, 28 May 2012 17:07:47 +0100 Message-ID: <4FC3A2D3.9000606@citrix.com> References: <1337855829-15683-1-git-send-email-david.vrabel@citrix.com> <1337855829-15683-2-git-send-email-david.vrabel@citrix.com> <7CE799CC0E4DE04B88D5FDF226E18AC2CDEB431490@LONPMAILBOX01.citrite.net> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <7CE799CC0E4DE04B88D5FDF226E18AC2CDEB431490@LONPMAILBOX01.citrite.net> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Frediano Ziglio Cc: George Dunlap , "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org On 28/05/12 17:03, Frediano Ziglio wrote: > On Thu, 2012-05-24 at 11:37 +0100, David Vrabel wrote: >> From: David Vrabel >> >> Trace hypercalls using a more useful trace record format. >> >> The EIP field is removed (it was always somewhere in the hypercall >> page) and include selected hypercall arguments (the number of calls in >> a multicall, and the number of PTE updates in an mmu_update). >> > > I think that EIP is quite useful as it allow to understand which code in > dom0 call that hypercall. The EIP was always an address in the hypercall page (i.e., hypercall_page + op * sizeof(unsigned long)) and doesn't tell you what made the hypercall. You would need one of the addresses off the guest stack to find the caller. > There is also space for an additional parameter without changing trace > version (adding information in a record should not be a problem). True, but George was keen on keeping the trace record size to a minimum. I am tempted to use 5 bits of the first extra word to indicate which parameters are present in the trace record. This might make the new format more future-proof, perhaps. David