From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:40732) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZCVef-0008Qn-VS for qemu-devel@nongnu.org; Tue, 07 Jul 2015 12:23:34 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZCVeb-00019R-Rj for qemu-devel@nongnu.org; Tue, 07 Jul 2015 12:23:33 -0400 Received: from greensocs.com ([193.104.36.180]:33467) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZCVeb-00018p-EQ for qemu-devel@nongnu.org; Tue, 07 Jul 2015 12:23:29 -0400 Message-ID: <559BFCFB.8070306@greensocs.com> Date: Tue, 07 Jul 2015 18:23:23 +0200 From: Frederic Konrad MIME-Version: 1.0 References: <1435330053-18733-1-git-send-email-fred.konrad@greensocs.com> <1435330053-18733-19-git-send-email-fred.konrad@greensocs.com> <878uas9c25.fsf@linaro.org> In-Reply-To: <878uas9c25.fsf@linaro.org> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [RFC PATCH V6 18/18] translate-all: (wip) use tb_flush_safe when we can't alloc more tb. List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?UTF-8?B?QWxleCBCZW5uw6ll?= Cc: mttcg@listserver.greensocs.com, peter.maydell@linaro.org, a.spyridakis@virtualopensystems.com, mark.burton@greensocs.com, agraf@suse.de, qemu-devel@nongnu.org, guillaume.delbergue@greensocs.com, pbonzini@redhat.com, alistair.francis@xilinx.com On 07/07/2015 18:17, Alex Benn=C3=A9e wrote: > fred.konrad@greensocs.com writes: > >> From: KONRAD Frederic >> >> This changes just the tb_flush called from tb_alloc. >> >> TODO: >> * changes the other tb_flush. >> >> Signed-off-by: KONRAD Frederic >> --- >> translate-all.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/translate-all.c b/translate-all.c >> index 8bd8fe8..9adaffa 100644 >> --- a/translate-all.c >> +++ b/translate-all.c >> @@ -1147,7 +1147,7 @@ TranslationBlock *tb_gen_code(CPUState *cpu, >> tb =3D tb_alloc(pc); >> if (!tb) { >> /* flush must be done */ >> - tb_flush(env); >> + tb_flush_safe(env); > Hold on this is async right? What stops us rolling on and then getting > flushed when the other vCPUs come to a halt? > > It deserves a comment at least. > not this need some synchronization when the CPUs are halted. There is crap here spotted by Paolo though. In the case of tb_flush.. We do an async_safe_work because all VCPUs thre= ad must be outside cpu_exec. Are you suggesting to just exiting everybody and wait here that all=20 VCPUs exit? This is possible as well here but is not possible for the other case..=20 That's why I prefered a "generic" mechanism. Fred >> /* cannot fail at this point */ >> tb =3D tb_alloc(pc); >> /* Don't forget to invalidate previous TB info. */