All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Emilio G. Cota" <cota@braap.org>
To: "Lluís Vilanova" <vilanova@ac.upc.edu>
Cc: qemu-devel@nongnu.org, Eric Blake <eblake@redhat.com>,
	Stefan Hajnoczi <stefanha@redhat.com>
Subject: Re: [Qemu-devel] [PATCH v4 11/20] instrument: Track vCPUs
Date: Wed, 6 Sep 2017 17:36:41 -0400	[thread overview]
Message-ID: <20170906213641.GE25558@flamenco> (raw)
In-Reply-To: <150472122675.24907.11597641982841030964.stgit@frigg.lan>

On Wed, Sep 06, 2017 at 21:07:06 +0300, Lluís Vilanova wrote:
> Keep a translation between instrumentation's QICPU and CPUState objects to avoid
> exposing QEMU's internals to instrumentation clients.
> 
> Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
(snip)
> diff --git a/instrument/control.c b/instrument/control.c
> index 2c2781beeb..83453ea561 100644
> --- a/instrument/control.c
> +++ b/instrument/control.c
> @@ -13,10 +13,32 @@
>  #include "instrument/load.h"
>  #include "instrument/qemu-instr/control.h"
>  #include "instrument/qemu-instr/visibility.h"
> +#include "qom/cpu.h"
> +
>  
>  __thread InstrState instr_cur_state;
>  
>  
> +unsigned int instr_cpus_count;
> +CPUState **instr_cpus;
> +
> +void instr_cpu_add(CPUState *vcpu)
> +{
> +    unsigned int idx = vcpu->cpu_index;
> +    if (idx >= instr_cpus_count) {
> +        instr_cpus_count = idx + 1;
> +        instr_cpus = realloc(instr_cpus, sizeof(*instr_cpus) * instr_cpus_count);
> +    }
> +    instr_cpus[idx] = vcpu;
> +}
> +
> +void instr_cpu_remove(CPUState *vcpu)
> +{
> +    unsigned int idx = vcpu->cpu_index;
> +    instr_cpus[idx] = NULL;
> +}

instr_cpus_count and instr_cpus are both modified with cpu_list_lock.
However, other readers do not acquire this same lock. This gets messy
when you try to implement something like "vcpu_for_each", which is
very useful when you load an instrumentation tool once the program
is running (otherwise the tool cannot know for sure what the vCPUs
are). This vcpu_for_each would also have to take cpu_list_lock, for
otherwise it'd miss new threads being added. As you can imagine this
gets messy pretty quickly, given that you have your own lock here.
I went with having my own hash table (a list would be fine too),
protected with the same "instr_lock" you have (i.e. the recursive mutex).

An unrelated comment: your usage of get/set/put confuses me. I'd expect those
to work on refcounted items; instead you use them for instance to hide a cast
(cpu_set) or to create/destroy (e.g. handle_get/put).

		E.

  reply	other threads:[~2017-09-06 21:36 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-09-06 17:22 [Qemu-devel] [PATCH v4 00/20] instrument: Add basic event instrumentation Lluís Vilanova
2017-09-06 17:26 ` [Qemu-devel] [PATCH v4 01/20] instrument: Add documentation Lluís Vilanova
2017-09-06 17:30 ` [Qemu-devel] [PATCH v4 02/20] instrument: Add configure-time flag Lluís Vilanova
2017-09-06 17:34 ` [Qemu-devel] [PATCH v4 03/20] instrument: Add generic library loader Lluís Vilanova
2017-09-06 21:20   ` Emilio G. Cota
2017-09-10 17:41     ` Lluís Vilanova
2017-09-06 17:38 ` [Qemu-devel] [PATCH v4 04/20] instrument: [linux-user] Add command line " Lluís Vilanova
2017-09-06 17:42 ` [Qemu-devel] [PATCH v4 05/20] instrument: [bsd-user] " Lluís Vilanova
2017-09-06 17:46 ` [Qemu-devel] [PATCH v4 06/20] instrument: [softmmu] " Lluís Vilanova
2017-09-06 17:50 ` [Qemu-devel] [PATCH v4 07/20] instrument: [qapi] Add " Lluís Vilanova
2017-09-07  8:43   ` Markus Armbruster
2017-09-10 21:39     ` Lluís Vilanova
2017-09-06 17:55 ` [Qemu-devel] [PATCH v4 08/20] instrument: [hmp] " Lluís Vilanova
2017-09-07  8:51   ` Markus Armbruster
2017-09-10 22:07     ` Lluís Vilanova
2017-09-06 17:59 ` [Qemu-devel] [PATCH v4 09/20] instrument: Add basic control interface Lluís Vilanova
2017-09-06 21:23   ` Emilio G. Cota
2017-09-10 22:15     ` Lluís Vilanova
2017-09-06 21:57   ` Emilio G. Cota
2017-09-10 23:28     ` Lluís Vilanova
2017-09-06 18:03 ` [Qemu-devel] [PATCH v4 10/20] instrument: Add support for tracing events Lluís Vilanova
2017-09-06 18:07 ` [Qemu-devel] [PATCH v4 11/20] instrument: Track vCPUs Lluís Vilanova
2017-09-06 21:36   ` Emilio G. Cota [this message]
2017-09-06 18:11 ` [Qemu-devel] [PATCH v4 12/20] instrument: Add event 'guest_cpu_enter' Lluís Vilanova
2017-09-06 18:15 ` [Qemu-devel] [PATCH v4 13/20] instrument: Add event 'guest_cpu_exit' Lluís Vilanova
2017-09-06 18:19 ` [Qemu-devel] [PATCH v4 14/20] instrument: Add event 'guest_cpu_reset' Lluís Vilanova
2017-09-06 18:23 ` [Qemu-devel] [PATCH v4 15/20] trace: Introduce a proper structure to describe memory accesses Lluís Vilanova
2017-09-06 18:27 ` [Qemu-devel] [PATCH v4 16/20] instrument: Add event 'guest_mem_before_trans' Lluís Vilanova
2017-09-06 18:31 ` [Qemu-devel] [PATCH v4 17/20] instrument: Add event 'guest_mem_before_exec' Lluís Vilanova
2017-09-06 18:35 ` [Qemu-devel] [PATCH v4 18/20] instrument: Add event 'guest_user_syscall' Lluís Vilanova
2017-09-06 18:39 ` [Qemu-devel] [PATCH v4 19/20] instrument: Add event 'guest_user_syscall_ret' Lluís Vilanova
2017-09-06 18:43 ` [Qemu-devel] [PATCH v4 20/20] instrument: Add API to manipulate guest memory Lluís Vilanova
2017-09-06 20:59 ` [Qemu-devel] [PATCH v4 00/20] instrument: Add basic event instrumentation Emilio G. Cota
2017-09-10 17:40   ` Lluís Vilanova
2017-09-10 23:31   ` Lluís Vilanova
2017-09-07  7:45 ` Markus Armbruster
2017-09-07 10:58 ` Markus Armbruster
2017-09-07 14:21   ` Emilio G. Cota
2017-09-10 17:34     ` Lluís Vilanova

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=20170906213641.GE25558@flamenco \
    --to=cota@braap.org \
    --cc=eblake@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.com \
    --cc=vilanova@ac.upc.edu \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.