From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:51787) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1cWLP4-0006np-OA for qemu-devel@nongnu.org; Wed, 25 Jan 2017 06:06:15 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1cWLP0-0005aD-7K for qemu-devel@nongnu.org; Wed, 25 Jan 2017 06:06:14 -0500 Received: from mx1.redhat.com ([209.132.183.28]:32786) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1cWLOz-0005ZZ-Vp for qemu-devel@nongnu.org; Wed, 25 Jan 2017 06:06:10 -0500 References: <20170124071654.4572.41407.stgit@PASHA-ISP> <20170124071719.4572.92155.stgit@PASHA-ISP> From: Paolo Bonzini Message-ID: Date: Wed, 25 Jan 2017 12:06:04 +0100 MIME-Version: 1.0 In-Reply-To: <20170124071719.4572.92155.stgit@PASHA-ISP> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH v7 04/14] icount: exit cpu loop on expire List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Pavel Dovgalyuk , qemu-devel@nongnu.org Cc: kwolf@redhat.com, peter.maydell@linaro.org, mst@redhat.com, jasowang@redhat.com, quintela@redhat.com, kraxel@redhat.com On 24/01/2017 08:17, Pavel Dovgalyuk wrote: > This patch adds check to break cpu loop when icount expires without > setting the TB_EXIT_ICOUNT_EXPIRED flag. It happens when there is no > available translated blocks and all instructions were executed. > In icount replay mode unnecessary tb_find will be called (which may > cause an exception) and execution will be non-deterministic. >=20 > Signed-off-by: Pavel Dovgalyuk > --- > cpu-exec.c | 15 ++++++++++++--- > 1 file changed, 12 insertions(+), 3 deletions(-) >=20 > diff --git a/cpu-exec.c b/cpu-exec.c > index 79a2167..e603da4 100644 > --- a/cpu-exec.c > +++ b/cpu-exec.c > @@ -582,9 +582,6 @@ static inline void cpu_loop_exec_tb(CPUState *cpu, = TranslationBlock *tb, > cpu_exec_nocache(cpu, insns_left, *last_tb, false); > align_clocks(sc, cpu); > } > - cpu->exception_index =3D EXCP_INTERRUPT; > - *last_tb =3D NULL; > - cpu_loop_exit(cpu); > } > break; > #endif > @@ -638,6 +635,18 @@ int cpu_exec(CPUState *cpu) > =20 > for(;;) { > cpu_handle_interrupt(cpu, &last_tb); > + /* icount has expired, we need to break the execution = loop. > + This check is needed before tb_find to make executi= on > + deterministic - tb_find may cause an exception > + while translating the code from non-mapped page. */ > + if (use_icount && ((cpu->icount_extra =3D=3D 0 > + && cpu->icount_decr.u16.low =3D=3D= 0) > + || (int32_t)cpu->icount_decr.u32 < 0))= { > + if (cpu->exception_index =3D=3D -1) { > + cpu->exception_index =3D EXCP_INTERRUPT; > + } > + cpu_loop_exit(cpu); > + } Can this can be placed in cpu_handle_interrupt itself? Also, can the cpu->exception_index !=3D -1 case happen? We really should add assertions on cpu->exception_index, maybe after cpu_handle_exception and cc->cpu_exec_interrupt returns true. Without some idea of the invariants, I'm not competent enough to review this patc= h. Paolo > tb =3D tb_find(cpu, last_tb, tb_exit); > cpu_loop_exec_tb(cpu, tb, &last_tb, &tb_exit, &sc); > /* Try to align the host and virtual clocks >=20