From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:60617) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dBbHT-00084w-6Z for qemu-devel@nongnu.org; Fri, 19 May 2017 02:20:57 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dBbHP-0006Qw-VY for qemu-devel@nongnu.org; Fri, 19 May 2017 02:20:55 -0400 Date: Fri, 19 May 2017 14:26:11 +1000 From: David Gibson Message-ID: <20170519042611.GE12284@umbus.fritz.box> References: <20170518215416.5613-1-danielhb@linux.vnet.ibm.com> <20170518215416.5613-2-danielhb@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="Rgf3q3z9SdmXC6oT" Content-Disposition: inline In-Reply-To: <20170518215416.5613-2-danielhb@linux.vnet.ibm.com> Subject: Re: [Qemu-devel] [RESEND PATCH v10 1/5] hw/ppc/spapr.c: adding pending_dimm_unplugs to sPAPRMachineState List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Daniel Henrique Barboza Cc: qemu-devel@nongnu.org, qemu-ppc@nongnu.org, mdroth@linux.vnet.ibm.com --Rgf3q3z9SdmXC6oT Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Thu, May 18, 2017 at 06:54:12PM -0300, Daniel Henrique Barboza wrote: > The LMB DRC release callback, spapr_lmb_release(), uses an opaque > parameter, a sPAPRDIMMState struct that stores the current LMBs that > are allocated to a DIMM (nr_lmbs). After each call to this callback, > the nr_lmbs is decremented by one and, when it reaches zero, the callback > proceeds with the qdev calls to hot unplug the LMB. >=20 > Using drc->detach_cb_opaque is problematic because it can't be migrated in > the future DRC migration work. This patch makes the following changes to > eliminate the usage of this opaque callback inside spapr_lmb_release: >=20 > - sPAPRDIMMState was moved from spapr.c and added to spapr.h. A new > attribute called 'addr' was added to it. This is used as an unique > identifier to associate a sPAPRDIMMState to a PCDIMM element. >=20 > - sPAPRMachineState now hosts a new QTAILQ called 'pending_dimm_unplugs'. > This queue of sPAPRDIMMState elements will store the DIMM state of DIMMs > that are currently going under an unplug process. >=20 > - spapr_lmb_release() will now retrieve the nr_lmbs value by getting the > correspondent sPAPRDIMMState. A helper function called spapr_dimm_get_add= ress > was created to fetch the address of a PCDIMM device inside spapr_lmb_rele= ase. > When nr_lmbs reaches zero and the callback proceeds with the qdev hot unp= lug > calls, the sPAPRDIMMState struct is removed from spapr->pending_dimm_unpl= ugs. >=20 > After these changes, the opaque argument for spapr_lmb_release is now > unused and is passed as NULL inside spapr_del_lmbs. This and the other > opaque arguments can now be safely removed from the code. >=20 > Signed-off-by: Daniel Henrique Barboza > --- > hw/ppc/spapr.c | 57 ++++++++++++++++++++++++++++++++++++++++++++= +----- > include/hw/ppc/spapr.h | 4 ++++ > 2 files changed, 56 insertions(+), 5 deletions(-) >=20 > diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c > index 0980d73..b05abe5 100644 > --- a/hw/ppc/spapr.c > +++ b/hw/ppc/spapr.c > @@ -2050,6 +2050,7 @@ static void ppc_spapr_init(MachineState *machine) > msi_nonbroken =3D true; > =20 > QLIST_INIT(&spapr->phbs); > + QTAILQ_INIT(&spapr->pending_dimm_unplugs); > =20 > /* Allocate RMA if necessary */ > rma_alloc_size =3D kvmppc_alloc_rma(&rma); > @@ -2603,20 +2604,63 @@ out: > error_propagate(errp, local_err); > } > =20 > -typedef struct sPAPRDIMMState { > +struct sPAPRDIMMState { > + uint64_t addr; Since you're not trying to migrate this any more, you can index the list by an actual PCDIMMDevice *, rather than the base address. You're already passing the DeviceState * for the DIMM around, so this will actually remove the address parameter from some functions. I think that could actually be done as a preliminary cleanup. It also probably makes sense to merge spapr_del_lmbs() with spapr_memory_unplug_request(), they're both very small. > uint32_t nr_lmbs; > -} sPAPRDIMMState; > + QTAILQ_ENTRY(sPAPRDIMMState) next; > +}; > + > +static sPAPRDIMMState *spapr_pending_dimm_unplugs_find(sPAPRMachineState= *s, > + uint64_t addr) > +{ > + sPAPRDIMMState *dimm_state =3D NULL; > + QTAILQ_FOREACH(dimm_state, &s->pending_dimm_unplugs, next) { > + if (dimm_state->addr =3D=3D addr) { > + break; > + } > + } > + return dimm_state; > +} > + > +static void spapr_pending_dimm_unplugs_add(sPAPRMachineState *spapr, > + sPAPRDIMMState *dimm_state) > +{ > + g_assert(!spapr_pending_dimm_unplugs_find(spapr, dimm_state->addr)); > + QTAILQ_INSERT_HEAD(&spapr->pending_dimm_unplugs, dimm_state, next); > +} > + > +static void spapr_pending_dimm_unplugs_remove(sPAPRMachineState *spapr, > + sPAPRDIMMState *dimm_state) > +{ > + QTAILQ_REMOVE(&spapr->pending_dimm_unplugs, dimm_state, next); > + g_free(dimm_state); > +} > + > +static uint64_t spapr_dimm_get_address(PCDIMMDevice *dimm) > +{ > + Error *local_err =3D NULL; > + uint64_t addr; > + addr =3D object_property_get_int(OBJECT(dimm), PC_DIMM_ADDR_PROP, > + &local_err); > + if (local_err) { > + error_propagate(&error_abort, local_err); > + return 0; > + } > + return addr; > +} > =20 > static void spapr_lmb_release(DeviceState *dev, void *opaque) > { > - sPAPRDIMMState *ds =3D (sPAPRDIMMState *)opaque; > HotplugHandler *hotplug_ctrl; > + uint64_t addr =3D spapr_dimm_get_address(PC_DIMM(dev)); > + sPAPRMachineState *spapr =3D SPAPR_MACHINE(qdev_get_machine()); I prefer not to access the machine as a global when possible. I think it's preferable to pass down the spapr object from above - unplug_request() itself can get it from hotplug_dev. > + sPAPRDIMMState *ds =3D spapr_pending_dimm_unplugs_find(spapr, addr); > =20 > if (--ds->nr_lmbs) { > return; > } > =20 > - g_free(ds); > + spapr_pending_dimm_unplugs_remove(spapr, ds); > =20 > /* > * Now that all the LMBs have been removed by the guest, call the > @@ -2633,17 +2677,20 @@ static void spapr_del_lmbs(DeviceState *dev, uint= 64_t addr_start, uint64_t size, > sPAPRDRConnectorClass *drck; > uint32_t nr_lmbs =3D size / SPAPR_MEMORY_BLOCK_SIZE; > int i; > + sPAPRMachineState *spapr =3D SPAPR_MACHINE(qdev_get_machine()); > sPAPRDIMMState *ds =3D g_malloc0(sizeof(sPAPRDIMMState)); > uint64_t addr =3D addr_start; > =20 > ds->nr_lmbs =3D nr_lmbs; > + ds->addr =3D addr_start; > + spapr_pending_dimm_unplugs_add(spapr, ds); > for (i =3D 0; i < nr_lmbs; i++) { > drc =3D spapr_dr_connector_by_id(SPAPR_DR_CONNECTOR_TYPE_LMB, > addr / SPAPR_MEMORY_BLOCK_SIZE); > g_assert(drc); > =20 > drck =3D SPAPR_DR_CONNECTOR_GET_CLASS(drc); > - drck->detach(drc, dev, spapr_lmb_release, ds, errp); > + drck->detach(drc, dev, spapr_lmb_release, NULL, errp); > addr +=3D SPAPR_MEMORY_BLOCK_SIZE; > } > =20 > diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h > index 5802f88..9823296 100644 > --- a/include/hw/ppc/spapr.h > +++ b/include/hw/ppc/spapr.h > @@ -32,6 +32,7 @@ struct sPAPRRTCState { > int64_t ns_offset; > }; > =20 > +typedef struct sPAPRDIMMState sPAPRDIMMState; > typedef struct sPAPRMachineClass sPAPRMachineClass; > =20 > #define TYPE_SPAPR_MACHINE "spapr-machine" > @@ -104,6 +105,9 @@ struct sPAPRMachineState { > /* RTAS state */ > QTAILQ_HEAD(, sPAPRConfigureConnectorState) ccs_list; > =20 > + /* pending DIMM unplug queue */ Perhaps update this to mention that it's a cache which can be regenerated when necessary. > + QTAILQ_HEAD(, sPAPRDIMMState) pending_dimm_unplugs; > + > /*< public >*/ > char *kvm_type; > MemoryHotplugState hotplug_memory; --=20 David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson --Rgf3q3z9SdmXC6oT Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBCAAGBQJZHnPhAAoJEGw4ysog2bOStz8QAMQ83MyfkyX6DgwbLxe05jHO K6e+VxopqqS6pP2XboAE9MIjI/ed0jsWFwnrp/vrM9GP7yeJyb+T0j2QFi+Wng+C LjD88GZ7KYEAeM04VFh7O5dqMzjhPtPFiBi+ka2a32qw3D0VEo12YWsq44roXW2G JabojeUCLO6PfQjZLsD0peab/DvXmmiInG//U3cmE68fNdLh/O6J7QzP2/vf31Vx ShO40oln1mr1SlrhjRj2lCJTUdiq4EXKIjlLfKBY0/L6X0DKuAtkjdKJsq0OuGVo PyfKd58LCBHf0f0nL4senecarn6QkiXqmqc02KM+FzpD5/hfdstePTrNtcWpLekt Vd+z9841XMbrc8aG3o4OO3V0HadJ4BoFrd7xn/zy1+SkMay9SAJWSSQIulwbCki0 fbCwzWLyM6vR9lXAit39B8fGYIFzXv49mLcbcO5lGmcapVJjEwTvnWD1sauFRHAU Tw80RIk86e1KlZ+ic9qGIlOHXtyVrl7yPR5xMMyh6h5rDNY+MJIKhGmgKjkGi1CC 4R6CzhRkD8JuRZqsjvAMqgiIWoq0BQRQXHd4MYT59NsmbJeSuThEtFrU7PBy9H3l pCiXXHhqEa0PJhPXamQk+Y+w3pkuT46Cr6+WkKHQYY3vJVLGSXy5B23XOZQKGzYZ djF9XFwo8h0bF2rB8mx5 =X24s -----END PGP SIGNATURE----- --Rgf3q3z9SdmXC6oT--