Date: Tue, 12 Mar 2019 09:24:58 +0000 (UTC) From: Kashyap D Desai <kadesai@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r345056 - head/sys/dev/mrsas Message-ID: <201903120924.x2C9Ow52017475@repo.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: kadesai Date: Tue Mar 12 09:24:58 2019 New Revision: 345056 URL: https://svnweb.freebsd.org/changeset/base/345056 Log: fw_outstanding"(outstanding IOs at firmware level) counter gets screwed up when R1 fastpath writes are running. Some of the cases which are not handled properly in driver are: 1. With R1 fastpath supported, single write from CAM layer can consume 2 MPT frames at driver/firmware level for fastpath qualification(if fw_outstanding < controller Queue Depth). Due to this driver has to throttle IOs coming from CAM layer as well as second fastpath write(of R1 write) against Adapter Queue Depth. If "fw_outstanding" reaches to adapter queue depth, driver should return IOs from CAM layer with device busy status.While allocating second MPT frame(corresponding to R1 FP write) also, driver should ensure fw_outstanding should not exceed adapter QD. 2. For R1 fastpath writes completion, driver decrements "fw_oustanding" counter without really returning MPT frame to free pool. It may cause IOs(with heavy IOs running, consuming whole adapter Queue Depth) consuming MPT frames reserved for DCMDs(management commands) and DCMDs(internal and sent by application) not getting MPT frame will start failing. Below is one test case to hit the issue described above- 1. Run heavy IOs (outstanding IOs should hit adapter Queue Depth). 2. Run management tool (Broadcom's storcli tool) querying adapter in loop (run command- "storcli64 /c0 show" in loop). 3. Management tool's requests would start failing due to non-availability of free MPT frames as all frames would be consumed by IOs. Fix: Increment/decrement of "fw_outstanding" counter should be in sync with MPT frame get/return. Submitted by: Sumit Saxena <sumit.saxena@broadcom.com> Reviewed by: Kashyap Desai <Kashyap.Desai@broadcom.com> Approved by: Ken MFC after: 3 days Sponsored by: Broadcom Inc Modified: head/sys/dev/mrsas/mrsas.c head/sys/dev/mrsas/mrsas_cam.c Modified: head/sys/dev/mrsas/mrsas.c ============================================================================== --- head/sys/dev/mrsas/mrsas.c Tue Mar 12 08:31:43 2019 (r345055) +++ head/sys/dev/mrsas/mrsas.c Tue Mar 12 09:24:58 2019 (r345056) @@ -1712,6 +1712,7 @@ mrsas_complete_cmd(struct mrsas_softc *sc, u_int32_t M mrsas_map_mpt_cmd_status(cmd_mpt, cmd_mpt->ccb_ptr, status, extStatus, data_length, sense); mrsas_cmd_done(sc, cmd_mpt); + mrsas_atomic_dec(&sc->fw_outstanding); } else { /* * If the peer Raid 1/10 fast path failed, @@ -1735,12 +1736,13 @@ mrsas_complete_cmd(struct mrsas_softc *sc, u_int32_t M r1_cmd->callout_owner = false; } mrsas_release_mpt_cmd(r1_cmd); + mrsas_atomic_dec(&sc->fw_outstanding); mrsas_map_mpt_cmd_status(cmd_mpt, cmd_mpt->ccb_ptr, status, extStatus, data_length, sense); mrsas_cmd_done(sc, cmd_mpt); + mrsas_atomic_dec(&sc->fw_outstanding); } } - mrsas_atomic_dec(&sc->fw_outstanding); break; case MRSAS_MPI2_FUNCTION_PASSTHRU_IO_REQUEST: /* MFI command */ cmd_mfi = sc->mfi_cmd_list[cmd_mpt->sync_cmd_idx]; @@ -2526,6 +2528,9 @@ mrsas_init_fw(struct mrsas_softc *sc) else sc->fast_path_io = 0; } + + device_printf(sc->mrsas_dev, "max_fw_cmds: %u max_scsi_cmds: %u\n", + sc->max_fw_cmds, sc->max_scsi_cmds); return (0); } Modified: head/sys/dev/mrsas/mrsas_cam.c ============================================================================== --- head/sys/dev/mrsas/mrsas_cam.c Tue Mar 12 08:31:43 2019 (r345055) +++ head/sys/dev/mrsas/mrsas_cam.c Tue Mar 12 09:24:58 2019 (r345056) @@ -467,11 +467,20 @@ mrsas_startio(struct mrsas_softc *sc, struct cam_sim * return (0); } ccb_h->status |= CAM_SIM_QUEUED; + + if (mrsas_atomic_inc_return(&sc->fw_outstanding) > sc->max_scsi_cmds) { + ccb_h->status |= CAM_REQUEUE_REQ; + xpt_done(ccb); + mrsas_atomic_dec(&sc->fw_outstanding); + return (0); + } + cmd = mrsas_get_mpt_cmd(sc); if (!cmd) { ccb_h->status |= CAM_REQUEUE_REQ; xpt_done(ccb); + mrsas_atomic_dec(&sc->fw_outstanding); return (0); } @@ -638,7 +647,7 @@ mrsas_startio(struct mrsas_softc *sc, struct cam_sim * mrsas_scsiio_timeout, cmd); #endif - if (mrsas_atomic_inc_return(&sc->fw_outstanding) > sc->io_cmds_highwater) + if (mrsas_atomic_read(&sc->fw_outstanding) > sc->io_cmds_highwater) sc->io_cmds_highwater++; /* @@ -653,7 +662,6 @@ mrsas_startio(struct mrsas_softc *sc, struct cam_sim * * new command */ if (cmd->r1_alt_dev_handle != MR_DEVHANDLE_INVALID) { - mrsas_atomic_inc(&sc->fw_outstanding); mrsas_prepare_secondRaid1_IO(sc, cmd); mrsas_fire_cmd(sc, req_desc->addr.u.low, req_desc->addr.u.high); @@ -669,6 +677,7 @@ mrsas_startio(struct mrsas_softc *sc, struct cam_sim * done: xpt_done(ccb); + mrsas_atomic_dec(&sc->fw_outstanding); return (0); } @@ -1092,14 +1101,20 @@ mrsas_setup_io(struct mrsas_softc *sc, struct mrsas_mp (io_info.r1_alt_dev_handle != MR_DEVHANDLE_INVALID) && (raid->level == 1) && !io_info.isRead) { r1_cmd = mrsas_get_mpt_cmd(sc); - if (!r1_cmd) { + if (mrsas_atomic_inc_return(&sc->fw_outstanding) > sc->max_scsi_cmds) { fp_possible = FALSE; - printf("Avago debug fp disable from %s %d \n", - __func__, __LINE__); + mrsas_atomic_dec(&sc->fw_outstanding); } else { - cmd->peer_cmd = r1_cmd; - r1_cmd->peer_cmd = cmd; - } + r1_cmd = mrsas_get_mpt_cmd(sc); + if (!r1_cmd) { + fp_possible = FALSE; + mrsas_atomic_dec(&sc->fw_outstanding); + } + else { + cmd->peer_cmd = r1_cmd; + r1_cmd->peer_cmd = cmd; + } + } } }
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201903120924.x2C9Ow52017475>