S: Maintained
F: drivers/hwmon/abituguru3.c
+ACCES 104-IDIO-16 GPIO DRIVER
+M: "William Breathitt Gray" <vilhelm.gray@gmail.com>
+L: linux-gpio@vger.kernel.org
+S: Maintained
+F: drivers/gpio/gpio-104-idio-16.c
+
ACENIC DRIVER
M: Jes Sorensen <jes@trained-monkey.org>
L: linux-acenic@sunsite.dk
F: drivers/gpu/drm/radeon/radeon_kfd.h
F: include/uapi/linux/kfd_ioctl.h
-AMD MICROCODE UPDATE SUPPORT
-M: Borislav Petkov <bp@alien8.de>
-S: Maintained
-F: arch/x86/kernel/cpu/microcode/amd*
-
AMD XGBE DRIVER
M: Tom Lendacky <thomas.lendacky@amd.com>
L: netdev@vger.kernel.org
F: drivers/net/appletalk/
F: net/appletalk/
+APPLIED MICRO (APM) X-GENE DEVICE TREE SUPPORT
+M: Duc Dang <dhdang@apm.com>
+S: Supported
+F: arch/arm64/boot/dts/apm/
+
APPLIED MICRO (APM) X-GENE SOC ETHERNET DRIVER
M: Iyappan Subramanian <isubramanian@apm.com>
M: Keyur Chudgar <kchudgar@apm.com>
ARM PMU PROFILING AND DEBUGGING
M: Will Deacon <will.deacon@arm.com>
+R: Mark Rutland <mark.rutland@arm.com>
S: Maintained
-F: arch/arm/kernel/perf_*
+F: arch/arm*/kernel/perf_*
F: arch/arm/oprofile/common.c
-F: arch/arm/kernel/hw_breakpoint.c
-F: arch/arm/include/asm/hw_breakpoint.h
-F: arch/arm/include/asm/perf_event.h
+F: arch/arm*/kernel/hw_breakpoint.c
+F: arch/arm*/include/asm/hw_breakpoint.h
+F: arch/arm*/include/asm/perf_event.h
F: drivers/perf/arm_pmu.c
F: include/linux/perf/arm_pmu.h
L: linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
S: Maintained
-ARM/Allwinner A1X SoC support
+ARM/Allwinner sunXi SoC support
M: Maxime Ripard <maxime.ripard@free-electrons.com>
+M: Chen-Yu Tsai <wens@csie.org>
L: linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
S: Maintained
-N: sun[x4567]i
+N: sun[x456789]i
ARM/Allwinner SoC Clock Support
M: Emilio López <emilio@elopez.com.ar>
S: Maintained
F: arch/arm/mach-alpine/
-ARM/ATMEL AT91RM9200 AND AT91SAM ARM ARCHITECTURES
+ARM/ATMEL AT91RM9200, AT91SAM9 AND SAMA5 SOC SUPPORT
M: Nicolas Ferre <nicolas.ferre@atmel.com>
M: Alexandre Belloni <alexandre.belloni@free-electrons.com>
M: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
M: Joachim Eastwood <manabian@gmail.com>
L: linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
S: Maintained
+F: arch/arm/boot/dts/lpc43*
+F: drivers/clk/nxp/clk-lpc18xx*
+F: drivers/clocksource/time-lpc32xx.c
+F: drivers/i2c/busses/i2c-lpc2k.c
+F: drivers/memory/pl172.c
+F: drivers/mtd/spi-nor/nxp-spifi.c
+F: drivers/rtc/rtc-lpc24xx.c
N: lpc18xx
ARM/MAGICIAN MACHINE SUPPORT
F: drivers/*/*s3c2410*
F: drivers/*/*/*s3c2410*
F: drivers/spi/spi-s3c*
+F: drivers/soc/samsung/*
F: sound/soc/samsung/*
+F: Documentation/arm/Samsung/
+F: Documentation/devicetree/bindings/arm/samsung/
+F: Documentation/devicetree/bindings/sram/samsung-sram.txt
+F: Documentation/devicetree/bindings/power/pd-samsung.txt
N: exynos
ARM/SAMSUNG MOBILE MACHINE SUPPORT
S: Maintained
F: drivers/media/platform/s5p-tv/
+ARM/SAMSUNG S5P SERIES JPEG CODEC SUPPORT
+M: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
+M: Jacek Anaszewski <j.anaszewski@samsung.com>
+L: linux-arm-kernel@lists.infradead.org
+L: linux-media@vger.kernel.org
+S: Maintained
+F: drivers/media/platform/s5p-jpeg/
+
ARM/SHMOBILE ARM ARCHITECTURE
M: Simon Horman <horms@verge.net.au>
M: Magnus Damm <magnus.damm@gmail.com>
F: arch/arm/boot/dts/r7s*
F: arch/arm/boot/dts/r8a*
F: arch/arm/boot/dts/sh*
-F: arch/arm/configs/bockw_defconfig
-F: arch/arm/configs/marzen_defconfig
F: arch/arm/configs/shmobile_defconfig
F: arch/arm/include/debug/renesas-scif.S
F: arch/arm/mach-shmobile/
S: Maintained
F: arch/arm/mach-sti/
F: arch/arm/boot/dts/sti*
+F: drivers/char/hw_random/st-rng.c
F: drivers/clocksource/arm_global_timer.c
F: drivers/clocksource/clksrc_st_lpc.c
F: drivers/i2c/busses/i2c-st.c
L: linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
S: Maintained
F: arch/arm/boot/dts/uniphier*
+F: arch/arm/include/asm/hardware/cache-uniphier.h
F: arch/arm/mach-uniphier/
+F: arch/arm/mm/cache-uniphier.c
+F: drivers/i2c/busses/i2c-uniphier*
F: drivers/pinctrl/uniphier/
F: drivers/tty/serial/8250/8250_uniphier.c
N: uniphier
F: Documentation/aoe/
F: drivers/block/aoe/
+ATHEROS 71XX/9XXX GPIO DRIVER
+M: Alban Bedel <albeu@free.fr>
+W: https://github.com/AlbanBedel/linux
+T: git git://github.com/AlbanBedel/linux
+S: Maintained
+F: drivers/gpio/gpio-ath79.c
+F: Documentation/devicetree/bindings/gpio/gpio-ath79.txt
+
ATHEROS ATH GENERIC UTILITIES
M: "Luis R. Rodriguez" <mcgrof@do-not-panic.com>
L: linux-wireless@vger.kernel.org
S: Supported
F: drivers/scsi/bnx2i/
-BROADCOM CYGNUS/IPROC ARM ARCHITECTURE
+BROADCOM IPROC ARM ARCHITECTURE
M: Ray Jui <rjui@broadcom.com>
M: Scott Branden <sbranden@broadcom.com>
+M: Jon Mason <jonmason@broadcom.com>
L: linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
L: bcm-kernel-feedback-list@broadcom.com
T: git git://github.com/broadcom/cygnus-linux.git
S: Maintained
N: iproc
N: cygnus
+N: nsp
N: bcm9113*
N: bcm9583*
-N: bcm583*
+N: bcm9585*
+N: bcm9586*
+N: bcm988312
N: bcm113*
+N: bcm583*
+N: bcm585*
+N: bcm586*
+N: bcm88312
BROADCOM BRCMSTB GPIO DRIVER
M: Gregory Fong <gregory.0xf0@gmail.com>
F: drivers/net/ethernet/cisco/enic/
CISCO VIC LOW LATENCY NIC DRIVER
- M: Upinder Malhi <umalhi@cisco.com>
+ M: Christian Benvenuti <benve@cisco.com>
+ M: Dave Goodell <dgoodell@cisco.com>
S: Supported
- F: drivers/infiniband/hw/usnic
+ F: drivers/infiniband/hw/usnic/
CIRRUS LOGIC EP93XX ETHERNET DRIVER
M: Hartley Sweeten <hsweeten@visionengravers.com>
W: http://www.dialog-semiconductor.com/products
S: Supported
F: Documentation/hwmon/da90??
+F: Documentation/devicetree/bindings/sound/da[79]*.txt
F: drivers/gpio/gpio-da90??.c
F: drivers/hwmon/da90??-hwmon.c
F: drivers/iio/adc/da91??-*.c
F: include/drm/i915*
F: include/uapi/drm/i915*
+DRM DRIVERS FOR ATMEL HLCDC
+M: Boris Brezillon <boris.brezillon@free-electrons.com>
+L: dri-devel@lists.freedesktop.org
+S: Supported
+F: drivers/gpu/drm/atmel-hlcdc/
+F: Documentation/devicetree/bindings/drm/atmel/
+
DRM DRIVERS FOR EXYNOS
M: Inki Dae <inki.dae@samsung.com>
M: Joonyoung Shim <jy0922.shim@samsung.com>
F: drivers/gpu/drm/imx/
F: Documentation/devicetree/bindings/drm/imx/
+DRM DRIVERS FOR GMA500 (Poulsbo, Moorestown and derivative chipsets)
+M: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
+L: dri-devel@lists.freedesktop.org
+T: git git://github.com/patjak/drm-gma500
+S: Maintained
+F: drivers/gpu/drm/gma500
+F: include/drm/gma500*
+
DRM DRIVERS FOR NVIDIA TEGRA
M: Thierry Reding <thierry.reding@gmail.com>
M: Terje Bergström <tbergstrom@nvidia.com>
F: sound/usb/misc/ua101.c
EXTENSIBLE FIRMWARE INTERFACE (EFI)
-M: Matt Fleming <matt.fleming@intel.com>
+M: Matt Fleming <matt@codeblueprint.co.uk>
L: linux-efi@vger.kernel.org
T: git git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git
S: Maintained
EFI VARIABLE FILESYSTEM
M: Matthew Garrett <matthew.garrett@nebula.com>
M: Jeremy Kerr <jk@ozlabs.org>
-M: Matt Fleming <matt.fleming@intel.com>
+M: Matt Fleming <matt@codeblueprint.co.uk>
T: git git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git
L: linux-efi@vger.kernel.org
S: Maintained
S: Maintained
F: drivers/net/ethernet/freescale/ucc_geth*
+FREESCALE eTSEC ETHERNET DRIVER (GIANFAR)
+M: Claudiu Manoil <claudiu.manoil@freescale.com>
+L: netdev@vger.kernel.org
+S: Maintained
+F: drivers/net/ethernet/freescale/gianfar*
+X: drivers/net/ethernet/freescale/gianfar_ptp.c
+F: Documentation/devicetree/bindings/net/fsl-tsec-phy.txt
+
FREESCALE QUICC ENGINE UCC UART DRIVER
M: Timur Tabi <timur@tabi.org>
L: linuxppc-dev@lists.ozlabs.org
F: Documentation/devicetree/bindings/i2c/
F: Documentation/i2c/
F: drivers/i2c/
+F: drivers/i2c/*/
F: include/linux/i2c.h
F: include/linux/i2c-*.h
F: include/uapi/linux/i2c.h
S: Supported
F: drivers/platform/x86/intel_menlow.c
-INTEL IA32 MICROCODE UPDATE SUPPORT
-M: Borislav Petkov <bp@alien8.de>
-S: Maintained
-F: arch/x86/kernel/cpu/microcode/core*
-F: arch/x86/kernel/cpu/microcode/intel*
-
INTEL I/OAT DMA DRIVER
M: Dave Jiang <dave.jiang@intel.com>
R: Dan Williams <dan.j.williams@intel.com>
INTEL WIRELESS WIFI LINK (iwlwifi)
M: Johannes Berg <johannes.berg@intel.com>
M: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
-M: Intel Linux Wireless <ilw@linux.intel.com>
+M: Intel Linux Wireless <linuxwifi@intel.com>
L: linux-wireless@vger.kernel.org
W: http://intellinuxwireless.org
T: git git://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi/iwlwifi.git
KERNEL VIRTUAL MACHINE (KVM) FOR AMD-V
M: Joerg Roedel <joro@8bytes.org>
L: kvm@vger.kernel.org
-W: http://kvm.qumranet.com
+W: http://www.linux-kvm.org/
S: Maintained
F: arch/x86/include/asm/svm.h
F: arch/x86/kvm/svm.c
KERNEL VIRTUAL MACHINE (KVM) FOR POWERPC
M: Alexander Graf <agraf@suse.com>
L: kvm-ppc@vger.kernel.org
-W: http://kvm.qumranet.com
+W: http://www.linux-kvm.org/
T: git git://github.com/agraf/linux-2.6.git
S: Supported
F: arch/powerpc/include/asm/kvm*
F: drivers/auxdisplay/ks0108.c
F: include/linux/ks0108.h
+L3MDEV
+M: David Ahern <dsa@cumulusnetworks.com>
+L: netdev@vger.kernel.org
+S: Maintained
+F: net/l3mdev
+F: include/net/l3mdev.h
+
LAPB module
L: linux-x25@vger.kernel.org
S: Orphan
S: Maintained
F: drivers/net/dsa/mv88e6352.c
+MARVELL CRYPTO DRIVER
+M: Boris Brezillon <boris.brezillon@free-electrons.com>
+M: Arnaud Ebalard <arno@natisbad.org>
+F: drivers/crypto/marvell/
+S: Maintained
+L: linux-crypto@vger.kernel.org
+
MARVELL GIGABIT ETHERNET DRIVERS (skge/sky2)
M: Mirko Lindner <mlindner@marvell.com>
M: Stephen Hemminger <stephen@networkplumber.org>
MELLANOX ETHERNET DRIVER (mlx4_en)
M: Amir Vadai <amirv@mellanox.com>
-M: Ido Shamay <idos@mellanox.com>
L: netdev@vger.kernel.org
S: Supported
W: http://www.mellanox.com
L: linux-wpan@vger.kernel.org
S: Maintained
F: drivers/net/ieee802154/mrf24j40.c
+F: Documentation/devicetree/bindings/net/ieee802154/mrf24j40.txt
MSI LAPTOP SUPPORT
M: "Lee, Chun-Yi" <jlee@suse.com>
F: drivers/net/
F: include/linux/if_*
F: include/linux/netdevice.h
-F: include/linux/arcdevice.h
F: include/linux/etherdevice.h
F: include/linux/fcdevice.h
F: include/linux/fddidevice.h
F: arch/x86/pci/
F: arch/x86/kernel/quirks.c
+PCI DRIVER FOR ALTERA PCIE IP
+M: Ley Foon Tan <lftan@altera.com>
+L: rfi@lists.rocketboards.org (moderated for non-subscribers)
+L: linux-pci@vger.kernel.org
+S: Supported
+F: Documentation/devicetree/bindings/pci/altera-pcie.txt
+F: drivers/pci/host/pcie-altera.c
+
PCI DRIVER FOR ARM VERSATILE PLATFORM
M: Rob Herring <robh@kernel.org>
L: linux-pci@vger.kernel.org
S: Maintained
F: drivers/pci/host/*spear*
+PCI MSI DRIVER FOR ALTERA MSI IP
+M: Ley Foon Tan <lftan@altera.com>
+L: rfi@lists.rocketboards.org (moderated for non-subscribers)
+L: linux-pci@vger.kernel.org
+S: Supported
+F: Documentation/devicetree/bindings/pci/altera-pcie-msi.txt
+F: drivers/pci/host/pcie-altera-msi.c
+
PCI MSI DRIVER FOR APPLIEDMICRO XGENE
M: Duc Dang <dhdang@apm.com>
L: linux-pci@vger.kernel.org
F: Documentation/devicetree/bindings/pci/xgene-pci-msi.txt
F: drivers/pci/host/pci-xgene-msi.c
+PCIE DRIVER FOR HISILICON
+M: Zhou Wang <wangzhou1@hisilicon.com>
+L: linux-pci@vger.kernel.org
+S: Maintained
+F: Documentation/devicetree/bindings/pci/hisilicon-pcie.txt
+F: drivers/pci/host/pcie-hisi.c
+
PCMCIA SUBSYSTEM
P: Linux PCMCIA Team
L: linux-pcmcia@lists.infradead.org
S: Maintained
F: drivers/pinctrl/pinctrl-at91.*
+PIN CONTROLLER - ATMEL AT91 PIO4
+M: Ludovic Desroches <ludovic.desroches@atmel.com>
+L: linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
+L: linux-gpio@vger.kernel.org
+S: Supported
+F: drivers/pinctrl/pinctrl-at91-pio4.*
+
PIN CONTROLLER - INTEL
M: Mika Westerberg <mika.westerberg@linux.intel.com>
M: Heikki Krogerus <heikki.krogerus@linux.intel.com>
S: Maintained
F: drivers/pnp/
-PNXxxxx I2C DRIVER
-M: Vitaly Wool <vitalywool@gmail.com>
-L: linux-i2c@vger.kernel.org
-S: Maintained
-F: drivers/i2c/busses/i2c-pnx.c
-
PPP PROTOCOL DRIVERS AND COMPRESSORS
M: Paul Mackerras <paulus@samba.org>
L: linux-ppp@vger.kernel.org
S: Supported
F: drivers/net/ethernet/qlogic/qlge/
+QLOGIC QL4xxx ETHERNET DRIVER
+M: Yuval Mintz <Yuval.Mintz@qlogic.com>
+M: Ariel Elior <Ariel.Elior@qlogic.com>
+M: everest-linux-l2@qlogic.com
+L: netdev@vger.kernel.org
+S: Supported
+F: drivers/net/ethernet/qlogic/qed/
+F: include/linux/qed/
+F: drivers/net/ethernet/qlogic/qede/
+
QNX4 FILESYSTEM
M: Anders Larsen <al@alarsen.net>
W: http://www.alarsen.net/linux/qnx4fs/
F: drivers/net/wireless/rtlwifi/
F: drivers/net/wireless/rtlwifi/rtl8192ce/
+RTL8XXXU WIRELESS DRIVER (rtl8xxxu)
+M: Jes Sorensen <Jes.Sorensen@redhat.com>
+L: linux-wireless@vger.kernel.org
+T: git git://git.kernel.org/pub/scm/linux/kernel/git/jes/linux.git rtl8723au-mac80211
+S: Maintained
+F: drivers/net/wireless/realtek/rtl8xxxu/
+
S3 SAVAGE FRAMEBUFFER DRIVER
M: Antonino Daplas <adaplas@gmail.com>
L: linux-fbdev@vger.kernel.org
F: Documentation/devicetree/bindings/net/snps,dwc-qos-ethernet.txt
F: drivers/net/ethernet/synopsys/dwc_eth_qos.c
+SYNOPSYS DESIGNWARE I2C DRIVER
+M: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
+M: Jarkko Nikula <jarkko.nikula@linux.intel.com>
+M: Mika Westerberg <mika.westerberg@linux.intel.com>
+L: linux-i2c@vger.kernel.org
+S: Maintained
+F: drivers/i2c/busses/i2c-designware-*
+F: include/linux/platform_data/i2c-designware.h
+
SYNOPSYS DESIGNWARE MMC/SD/SDIO DRIVER
M: Seungwon Jeon <tgih.jun@samsung.com>
M: Jaehoon Chung <jh80.chung@samsung.com>
S: Supported
F: arch/score/
+SYSTEM CONTROL & POWER INTERFACE (SCPI) Message Protocol drivers
+M: Sudeep Holla <sudeep.holla@arm.com>
+L: linux-arm-kernel@lists.infradead.org
+S: Maintained
+F: Documentation/devicetree/bindings/arm/arm,scpi.txt
+F: drivers/clk/clk-scpi.c
+F: drivers/cpufreq/scpi-cpufreq.c
+F: drivers/firmware/arm_scpi.c
+F: include/linux/scpi_protocol.h
+
SCSI CDROM DRIVER
M: Jens Axboe <axboe@kernel.dk>
L: linux-scsi@vger.kernel.org
F: drivers/staging/lustre
STAGING - NVIDIA COMPLIANT EMBEDDED CONTROLLER INTERFACE (nvec)
-M: Julian Andres Klode <jak@jak-linux.org>
M: Marc Dietrich <marvin24@gmx.de>
L: ac100@lists.launchpad.net (moderated for non-subscribers)
L: linux-tegra@vger.kernel.org
SYNOPSYS ARC ARCHITECTURE
M: Vineet Gupta <vgupta@synopsys.com>
+L: linux-snps-arc@lists.infraded.org
S: Supported
F: arch/arc/
F: Documentation/devicetree/bindings/arc/*
F: drivers/media/v4l2-core/videobuf2-*
F: include/media/videobuf2-*
+VIRTUAL SERIO DEVICE DRIVER
+M: Stephen Chandler Paul <thatslyude@gmail.com>
+S: Maintained
+F: drivers/input/serio/userio.c
+F: include/uapi/linux/userio.h
+
VIRTIO CONSOLE DRIVER
M: Amit Shah <amit.shah@redhat.com>
L: virtualization@lists.linux-foundation.org
L: netdev@vger.kernel.org
S: Maintained
F: drivers/net/vrf.c
-F: include/net/vrf.h
F: Documentation/networking/vrf.txt
VT1211 HARDWARE MONITOR DRIVER
S: Maintained
F: drivers/net/wireless/wl3501*
-WM97XX TOUCHSCREEN DRIVERS
-M: Mark Brown <broonie@kernel.org>
-M: Liam Girdwood <lrg@slimlogic.co.uk>
-L: linux-input@vger.kernel.org
-W: https://github.com/CirrusLogic/linux-drivers/wiki
-S: Supported
-F: drivers/input/touchscreen/*wm97*
-F: include/linux/wm97xx.h
-
WOLFSON MICROELECTRONICS DRIVERS
L: patches@opensource.wolfsonmicro.com
T: git https://github.com/CirrusLogic/linux-drivers.git
S: Maintained
F: arch/x86/kernel/cpu/mcheck/*
+X86 MICROCODE UPDATE SUPPORT
+M: Borislav Petkov <bp@alien8.de>
+S: Maintained
+F: arch/x86/kernel/cpu/microcode/*
+
X86 VDSO
M: Andy Lutomirski <luto@amacapital.net>
L: linux-kernel@vger.kernel.org
ZSMALLOC COMPRESSED SLAB MEMORY ALLOCATOR
M: Minchan Kim <minchan@kernel.org>
M: Nitin Gupta <ngupta@vflare.org>
+R: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
L: linux-mm@kvack.org
S: Maintained
F: mm/zsmalloc.c
* SOFTWARE.
*/
-#include <asm-generic/kmap_types.h>
+#include <linux/highmem.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/errno.h>
dev->ib_dev.detach_mcast = mlx5_ib_mcg_detach;
dev->ib_dev.process_mad = mlx5_ib_process_mad;
dev->ib_dev.alloc_mr = mlx5_ib_alloc_mr;
- dev->ib_dev.alloc_fast_reg_page_list = mlx5_ib_alloc_fast_reg_page_list;
- dev->ib_dev.free_fast_reg_page_list = mlx5_ib_free_fast_reg_page_list;
+ dev->ib_dev.map_mr_sg = mlx5_ib_map_mr_sg;
dev->ib_dev.check_mr_status = mlx5_ib_check_mr_status;
dev->ib_dev.get_port_immutable = mlx5_port_immutable;
#include <linux/timer.h>
#include <linux/vmalloc.h>
#include <linux/etherdevice.h>
+#include <linux/net_tstamp.h>
#include <asm/io.h>
+ #include "t4_chip_type.h"
#include "cxgb4_uld.h"
#define CH_WARN(adap, fmt, ...) dev_warn(adap->pdev_dev, fmt, ## __VA_ARGS__)
unsigned char width;
};
- #define CHELSIO_CHIP_CODE(version, revision) (((version) << 4) | (revision))
- #define CHELSIO_CHIP_FPGA 0x100
- #define CHELSIO_CHIP_VERSION(code) (((code) >> 4) & 0xf)
- #define CHELSIO_CHIP_RELEASE(code) ((code) & 0xf)
-
- #define CHELSIO_T4 0x4
- #define CHELSIO_T5 0x5
- #define CHELSIO_T6 0x6
-
- enum chip_type {
- T4_A1 = CHELSIO_CHIP_CODE(CHELSIO_T4, 1),
- T4_A2 = CHELSIO_CHIP_CODE(CHELSIO_T4, 2),
- T4_FIRST_REV = T4_A1,
- T4_LAST_REV = T4_A2,
-
- T5_A0 = CHELSIO_CHIP_CODE(CHELSIO_T5, 0),
- T5_A1 = CHELSIO_CHIP_CODE(CHELSIO_T5, 1),
- T5_FIRST_REV = T5_A0,
- T5_LAST_REV = T5_A1,
-
- T6_A0 = CHELSIO_CHIP_CODE(CHELSIO_T6, 0),
- T6_FIRST_REV = T6_A0,
- T6_LAST_REV = T6_A0,
- };
-
struct devlog_params {
u32 memtype; /* which memory (EDC0, EDC1, MC) */
u32 start; /* start of log in firmware memory */
#ifdef CONFIG_CHELSIO_T4_FCOE
struct cxgb_fcoe fcoe;
#endif /* CONFIG_CHELSIO_T4_FCOE */
+ bool rxtstamp; /* Enable TS */
+ struct hwtstamp_config tstamp_config;
};
struct dentry;
/* A packet gather list */
struct pkt_gl {
+ u64 sgetstamp; /* SGE Time Stamp for Ingress Packet */
struct page_frag frags[MAX_SKB_FRAGS];
void *va; /* virtual address of first byte */
unsigned int nfrags; /* # of fragments */
return adap->params.offload;
}
- static inline int is_t6(enum chip_type chip)
- {
- return CHELSIO_CHIP_VERSION(chip) == CHELSIO_T6;
- }
-
- static inline int is_t5(enum chip_type chip)
- {
- return CHELSIO_CHIP_VERSION(chip) == CHELSIO_T5;
- }
-
- static inline int is_t4(enum chip_type chip)
- {
- return CHELSIO_CHIP_VERSION(chip) == CHELSIO_T4;
- }
-
static inline u32 t4_read_reg(struct adapter *adap, u32 reg_addr)
{
return readl(adap->regs + reg_addr);
#endif
#define DRV_VERSION "2.0.0-ko"
const char cxgb4_driver_version[] = DRV_VERSION;
-#define DRV_DESC "Chelsio T4/T5 Network Driver"
+#define DRV_DESC "Chelsio T4/T5/T6 Network Driver"
/* Host shadow copy of ingress filter entry. This is in host native format
* and doesn't match the ordering or bit order, etc. of the hardware of the
MODULE_DEVICE_TABLE(pci, cxgb4_pci_tbl);
MODULE_FIRMWARE(FW4_FNAME);
MODULE_FIRMWARE(FW5_FNAME);
+MODULE_FIRMWARE(FW6_FNAME);
/*
* Normally we're willing to become the firmware's Master PF but will be happy
else {
static const char *fc[] = { "no", "Rx", "Tx", "Tx/Rx" };
- const char *s = "10Mbps";
+ const char *s;
const struct port_info *p = netdev_priv(dev);
switch (p->link_cfg.speed) {
case 40000:
s = "40Gbps";
break;
+ default:
+ pr_info("%s: unsupported speed: %d\n",
+ dev->name, p->link_cfg.speed);
+ return;
}
netdev_info(dev, "link up, %s, full-duplex, %s PAUSE\n", s,
}
EXPORT_SYMBOL(cxgb4_best_aligned_mtu);
+ /**
+ * cxgb4_tp_smt_idx - Get the Source Mac Table index for this VI
+ * @chip: chip type
+ * @viid: VI id of the given port
+ *
+ * Return the SMT index for this VI.
+ */
+ unsigned int cxgb4_tp_smt_idx(enum chip_type chip, unsigned int viid)
+ {
+ /* In T4/T5, SMT contains 256 SMAC entries organized in
+ * 128 rows of 2 entries each.
+ * In T6, SMT contains 256 SMAC entries in 256 rows.
+ * TODO: The below code needs to be updated when we add support
+ * for 256 VFs.
+ */
+ if (CHELSIO_CHIP_VERSION(chip) <= CHELSIO_T5)
+ return ((viid & 0x7f) << 1);
+ else
+ return (viid & 0x7f);
+ }
+ EXPORT_SYMBOL(cxgb4_tp_smt_idx);
+
/**
* cxgb4_port_chan - get the HW channel of a port
* @dev: the net device for the port
ret = t4_mdio_wr(pi->adapter, mbox, prtad, devad,
data->reg_num, data->val_in);
break;
+ case SIOCGHWTSTAMP:
+ return copy_to_user(req->ifr_data, &pi->tstamp_config,
+ sizeof(pi->tstamp_config)) ?
+ -EFAULT : 0;
+ case SIOCSHWTSTAMP:
+ if (copy_from_user(&pi->tstamp_config, req->ifr_data,
+ sizeof(pi->tstamp_config)))
+ return -EFAULT;
+
+ switch (pi->tstamp_config.rx_filter) {
+ case HWTSTAMP_FILTER_NONE:
+ pi->rxtstamp = false;
+ break;
+ case HWTSTAMP_FILTER_ALL:
+ pi->rxtstamp = true;
+ break;
+ default:
+ pi->tstamp_config.rx_filter = HWTSTAMP_FILTER_NONE;
+ return -ERANGE;
+ }
+
+ return copy_to_user(req->ifr_data, &pi->tstamp_config,
+ sizeof(pi->tstamp_config)) ?
+ -EFAULT : 0;
default:
return -EOPNOTSUPP;
}
t4_get_tp_version(adap, &adap->params.tp_vers);
ret = t4_check_fw_version(adap);
/* If firmware is too old (not supported by driver) force an update. */
- if (ret == -EFAULT)
+ if (ret)
state = DEV_STATE_UNINIT;
if ((adap->flags & MASTER_PF) && state != DEV_STATE_INIT) {
struct fw_info *fw_info;
}
for (i = 0; i < allocated; ++i)
adap->msix_info[i].vec = entries[i].vector;
+ dev_info(adap->pdev_dev, "%d MSI-X vectors allocated, "
+ "nic %d iscsi %d rdma cpl %d rdma ciq %d\n",
+ allocated, s->max_ethqsets, s->ofldqsets, s->rdmaqs,
+ s->rdmaciqs);
kfree(entries);
return 0;
[27] = "Port beacon support",
[28] = "RX-ALL support",
[29] = "802.1ad offload support",
+ [31] = "Modifying loopback source checks using UPDATE_QP support",
+ [32] = "Loopback source checks support",
};
int i;
MLX4_GET(field32, outbox, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET);
if (field32 & (1 << 16))
dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_UPDATE_QP;
+ if (field32 & (1 << 18))
+ dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB;
+ if (field32 & (1 << 19))
+ dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_LB_SRC_CHK;
if (field32 & (1 << 26))
dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_VLAN_CONTROL;
if (field32 & (1 << 20))
return -EOPNOTSUPP;
}
EXPORT_SYMBOL(set_phv_bit);
+
+void mlx4_replace_zero_macs(struct mlx4_dev *dev)
+{
+ int i;
+ u8 mac_addr[ETH_ALEN];
+
+ dev->port_random_macs = 0;
+ for (i = 1; i <= dev->caps.num_ports; ++i)
+ if (!dev->caps.def_mac[i] &&
+ dev->caps.port_type[i] == MLX4_PORT_TYPE_ETH) {
+ eth_random_addr(mac_addr);
+ dev->port_random_macs |= 1 << i;
+ dev->caps.def_mac[i] = mlx4_mac_to_u64(mac_addr);
+ }
+}
+EXPORT_SYMBOL_GPL(mlx4_replace_zero_macs);
u64 qp_mask = 0;
int err = 0;
+ if (!attr || (attr & ~MLX4_UPDATE_QP_SUPPORTED_ATTRS))
+ return -EINVAL;
+
mailbox = mlx4_alloc_cmd_mailbox(dev);
if (IS_ERR(mailbox))
return PTR_ERR(mailbox);
cmd = (struct mlx4_update_qp_context *)mailbox->buf;
- if (!attr || (attr & ~MLX4_UPDATE_QP_SUPPORTED_ATTRS))
- return -EINVAL;
-
if (attr & MLX4_UPDATE_QP_SMAC) {
pri_addr_path_mask |= 1ULL << MLX4_UPD_QP_PATH_MASK_MAC_INDEX;
cmd->qp_context.pri_path.grh_mylmc = params->smac_index;
}
+ if (attr & MLX4_UPDATE_QP_ETH_SRC_CHECK_MC_LB) {
+ if (!(dev->caps.flags2
+ & MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB)) {
+ mlx4_warn(dev,
+ "Trying to set src check LB, but it isn't supported\n");
+ err = -ENOTSUPP;
+ goto out;
+ }
+ pri_addr_path_mask |=
+ 1ULL << MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_MC_LB;
+ if (params->flags &
+ MLX4_UPDATE_QP_PARAMS_FLAGS_ETH_CHECK_MC_LB) {
+ cmd->qp_context.pri_path.fl |=
+ MLX4_FL_ETH_SRC_CHECK_MC_LB;
+ }
+ }
+
if (attr & MLX4_UPDATE_QP_VSD) {
qp_mask |= 1ULL << MLX4_UPD_QP_MASK_VSD;
if (params->flags & MLX4_UPDATE_QP_PARAMS_FLAGS_VSD_ENABLE)
err = mlx4_cmd(dev, mailbox->dma, qpn & 0xffffff, 0,
MLX4_CMD_UPDATE_QP, MLX4_CMD_TIME_CLASS_A,
MLX4_CMD_NATIVE);
-
+ out:
mlx4_free_cmd_mailbox(dev, mailbox);
return err;
}
}
}
+ /* preserve IF_COUNTER flag */
+ qpc->pri_path.vlan_control &=
+ MLX4_CTRL_ETH_SRC_CHECK_IF_COUNTER;
if (vp_oper->state.link_state == IFLA_VF_LINK_STATE_DISABLE &&
dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP) {
- qpc->pri_path.vlan_control =
+ qpc->pri_path.vlan_control |=
MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED |
MLX4_VLAN_CTRL_ETH_TX_BLOCK_UNTAGGED |
MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED |
MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
} else if (0 != vp_oper->state.default_vlan) {
- qpc->pri_path.vlan_control =
+ qpc->pri_path.vlan_control |=
MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED;
} else { /* priority tagged */
- qpc->pri_path.vlan_control =
+ qpc->pri_path.vlan_control |=
MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
}
return 0;
undo:
- for (--i; i >= base; --i)
+ for (--i; i >= 0; --i) {
rb_erase(&res_arr[i]->node, root);
+ list_del_init(&res_arr[i]->list);
+ }
spin_unlock_irq(mlx4_tlock(dev));
update_gid(dev, inbox, (u8)slave);
adjust_proxy_tun_qkey(dev, vhcr, qpc);
orig_sched_queue = qpc->pri_path.sched_queue;
- err = update_vport_qp_param(dev, inbox, slave, qpn);
- if (err)
- return err;
err = get_res(dev, slave, qpn, RES_QP, &qp);
if (err)
goto out;
}
+ err = update_vport_qp_param(dev, inbox, slave, qpn);
+ if (err)
+ goto out;
+
err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
out:
/* if no error, save sched queue value passed in by VF. This is
}
- #define MLX4_UPD_QP_PATH_MASK_SUPPORTED (1ULL << MLX4_UPD_QP_PATH_MASK_MAC_INDEX)
+ #define MLX4_UPD_QP_PATH_MASK_SUPPORTED ( \
+ 1ULL << MLX4_UPD_QP_PATH_MASK_MAC_INDEX |\
+ 1ULL << MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_MC_LB)
int mlx4_UPDATE_QP_wrapper(struct mlx4_dev *dev, int slave,
struct mlx4_vhcr *vhcr,
struct mlx4_cmd_mailbox *inbox,
(pri_addr_path_mask & ~MLX4_UPD_QP_PATH_MASK_SUPPORTED))
return -EPERM;
+ if ((pri_addr_path_mask &
+ (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_MC_LB)) &&
+ !(dev->caps.flags2 &
+ MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB)) {
+ mlx4_warn(dev,
+ "Src check LB for slave %d isn't supported\n",
+ slave);
+ return -ENOTSUPP;
+ }
+
/* Just change the smac for the QP */
err = get_res(dev, slave, qpn, RES_QP, &rqp);
if (err) {
MLX4_DEV_CAP_FLAG2_IGNORE_FCS = 1LL << 28,
MLX4_DEV_CAP_FLAG2_PHV_EN = 1LL << 29,
MLX4_DEV_CAP_FLAG2_SKIP_OUTER_VLAN = 1LL << 30,
+ MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB = 1ULL << 31,
+ MLX4_DEV_CAP_FLAG2_LB_SRC_CHK = 1ULL << 32,
};
enum {
struct mlx4_quotas quotas;
struct radix_tree_root qp_table_tree;
u8 rev_id;
+ u8 port_random_macs;
char board_id[MLX4_BOARD_ID_LEN];
int numa_node;
int oper_log_mgm_entry_size;
};
struct svc_rdma_fastreg_mr {
struct ib_mr *mr;
- void *kva;
- struct ib_fast_reg_page_list *page_list;
- int page_list_len;
+ struct scatterlist *sg;
+ int sg_nents;
unsigned long access_flags;
- unsigned long map_len;
enum dma_data_direction direction;
struct list_head frmr_list;
};
struct svc_rdma_fastreg_mr *);
extern void svc_sq_reap(struct svcxprt_rdma *);
extern void svc_rq_reap(struct svcxprt_rdma *);
-extern struct svc_xprt_class svc_rdma_class;
extern void svc_rdma_prep_reply_hdr(struct svc_rqst *);
+extern struct svc_xprt_class svc_rdma_class;
+#ifdef CONFIG_SUNRPC_BACKCHANNEL
+extern struct svc_xprt_class svc_rdma_bc_class;
+#endif
+
/* svc_rdma.c */
extern int svc_rdma_init(void);
extern void svc_rdma_cleanup(void);
#include "rds.h"
#include "ib.h"
-static unsigned int fmr_pool_size = RDS_FMR_POOL_SIZE;
-unsigned int fmr_message_size = RDS_FMR_SIZE + 1; /* +1 allows for unaligned MRs */
+unsigned int rds_ib_fmr_1m_pool_size = RDS_FMR_1M_POOL_SIZE;
+unsigned int rds_ib_fmr_8k_pool_size = RDS_FMR_8K_POOL_SIZE;
unsigned int rds_ib_retry_count = RDS_IB_DEFAULT_RETRY_COUNT;
-module_param(fmr_pool_size, int, 0444);
-MODULE_PARM_DESC(fmr_pool_size, " Max number of fmr per HCA");
-module_param(fmr_message_size, int, 0444);
-MODULE_PARM_DESC(fmr_message_size, " Max size of a RDMA transfer");
+module_param(rds_ib_fmr_1m_pool_size, int, 0444);
+MODULE_PARM_DESC(rds_ib_fmr_1m_pool_size, " Max number of 1M fmr per HCA");
+module_param(rds_ib_fmr_8k_pool_size, int, 0444);
+MODULE_PARM_DESC(rds_ib_fmr_8k_pool_size, " Max number of 8K fmr per HCA");
module_param(rds_ib_retry_count, int, 0444);
MODULE_PARM_DESC(rds_ib_retry_count, " Number of hw retries before reporting an error");
struct rds_ib_device *rds_ibdev = container_of(work,
struct rds_ib_device, free_work);
- if (rds_ibdev->mr_pool)
- rds_ib_destroy_mr_pool(rds_ibdev->mr_pool);
+ if (rds_ibdev->mr_8k_pool)
+ rds_ib_destroy_mr_pool(rds_ibdev->mr_8k_pool);
+ if (rds_ibdev->mr_1m_pool)
+ rds_ib_destroy_mr_pool(rds_ibdev->mr_1m_pool);
if (rds_ibdev->pd)
ib_dealloc_pd(rds_ibdev->pd);
rds_ibdev->max_sge = min(dev_attr->max_sge, RDS_IB_MAX_SGE);
rds_ibdev->fmr_max_remaps = dev_attr->max_map_per_fmr?: 32;
- rds_ibdev->max_fmrs = dev_attr->max_fmr ?
- min_t(unsigned int, dev_attr->max_fmr, fmr_pool_size) :
- fmr_pool_size;
+ rds_ibdev->max_1m_fmrs = dev_attr->max_mr ?
+ min_t(unsigned int, (dev_attr->max_mr / 2),
+ rds_ib_fmr_1m_pool_size) : rds_ib_fmr_1m_pool_size;
+
+ rds_ibdev->max_8k_fmrs = dev_attr->max_mr ?
+ min_t(unsigned int, ((dev_attr->max_mr / 2) * RDS_MR_8K_SCALE),
+ rds_ib_fmr_8k_pool_size) : rds_ib_fmr_8k_pool_size;
rds_ibdev->max_initiator_depth = dev_attr->max_qp_init_rd_atom;
rds_ibdev->max_responder_resources = dev_attr->max_qp_rd_atom;
goto put_dev;
}
- rds_ibdev->mr_pool = rds_ib_create_mr_pool(rds_ibdev);
- if (IS_ERR(rds_ibdev->mr_pool)) {
- rds_ibdev->mr_pool = NULL;
+ rds_ibdev->mr_1m_pool =
+ rds_ib_create_mr_pool(rds_ibdev, RDS_IB_MR_1M_POOL);
+ if (IS_ERR(rds_ibdev->mr_1m_pool)) {
+ rds_ibdev->mr_1m_pool = NULL;
goto put_dev;
}
+ rds_ibdev->mr_8k_pool =
+ rds_ib_create_mr_pool(rds_ibdev, RDS_IB_MR_8K_POOL);
+ if (IS_ERR(rds_ibdev->mr_8k_pool)) {
+ rds_ibdev->mr_8k_pool = NULL;
+ goto put_dev;
+ }
+
+ rdsdebug("RDS/IB: max_mr = %d, max_wrs = %d, max_sge = %d, fmr_max_remaps = %d, max_1m_fmrs = %d, max_8k_fmrs = %d\n",
+ dev_attr->max_fmr, rds_ibdev->max_wrs, rds_ibdev->max_sge,
+ rds_ibdev->fmr_max_remaps, rds_ibdev->max_1m_fmrs,
+ rds_ibdev->max_8k_fmrs);
+
INIT_LIST_HEAD(&rds_ibdev->ipaddr_list);
INIT_LIST_HEAD(&rds_ibdev->conn_list);
/* Create a CMA ID and try to bind it. This catches both
* IB and iWARP capable NICs.
*/
- cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP, IB_QPT_RC);
+ cm_id = rdma_create_id(&init_net, NULL, NULL, RDMA_PS_TCP, IB_QPT_RC);
if (IS_ERR(cm_id))
return PTR_ERR(cm_id);
#include "rds.h"
#include "rdma_transport.h"
-#define RDS_FMR_SIZE 256
-#define RDS_FMR_POOL_SIZE 8192
+#define RDS_FMR_1M_POOL_SIZE (8192 / 2)
+#define RDS_FMR_1M_MSG_SIZE 256
+#define RDS_FMR_8K_MSG_SIZE 2
+#define RDS_MR_8K_SCALE (256 / (RDS_FMR_8K_MSG_SIZE + 1))
+#define RDS_FMR_8K_POOL_SIZE (RDS_MR_8K_SCALE * (8192 / 2))
#define RDS_IB_MAX_SGE 8
#define RDS_IB_RECV_SGE 2
#define RDS_IB_RECYCLE_BATCH_COUNT 32
+#define RDS_IB_WC_MAX 32
+#define RDS_IB_SEND_OP BIT_ULL(63)
+
extern struct rw_semaphore rds_ib_devices_lock;
extern struct list_head rds_ib_devices;
struct rds_ib_send_work {
void *s_op;
- struct ib_send_wr s_wr;
+ union {
+ struct ib_send_wr s_wr;
+ struct ib_rdma_wr s_rdma_wr;
+ struct ib_atomic_wr s_atomic_wr;
+ };
struct ib_sge s_sge[RDS_IB_MAX_SGE];
unsigned long s_queued;
};
atomic_t w_free_ctr;
};
+/* Rings are posted with all the allocations they'll need to queue the
+ * incoming message to the receiving socket so this can't fail.
+ * All fragments start with a header, so we can make sure we're not receiving
+ * garbage, and we can tell a small 8 byte fragment from an ACK frame.
+ */
+struct rds_ib_ack_state {
+ u64 ack_next;
+ u64 ack_recv;
+ unsigned int ack_required:1;
+ unsigned int ack_next_valid:1;
+ unsigned int ack_recv_valid:1;
+};
+
+
struct rds_ib_device;
struct rds_ib_connection {
struct ib_pd *i_pd;
struct ib_cq *i_send_cq;
struct ib_cq *i_recv_cq;
+ struct ib_wc i_send_wc[RDS_IB_WC_MAX];
+ struct ib_wc i_recv_wc[RDS_IB_WC_MAX];
+
+ /* interrupt handling */
+ struct tasklet_struct i_send_tasklet;
+ struct tasklet_struct i_recv_tasklet;
/* tx */
struct rds_ib_work_ring i_send_ring;
atomic_t i_signaled_sends;
/* rx */
- struct tasklet_struct i_recv_tasklet;
struct mutex i_recv_mutex;
struct rds_ib_work_ring i_recv_ring;
struct rds_ib_incoming *i_ibinc;
struct rds_ib_ipaddr {
struct list_head list;
__be32 ipaddr;
+ struct rcu_head rcu;
+};
+
+enum {
+ RDS_IB_MR_8K_POOL,
+ RDS_IB_MR_1M_POOL,
};
struct rds_ib_device {
struct list_head conn_list;
struct ib_device *dev;
struct ib_pd *pd;
- struct rds_ib_mr_pool *mr_pool;
- unsigned int fmr_max_remaps;
unsigned int max_fmrs;
+ struct rds_ib_mr_pool *mr_1m_pool;
+ struct rds_ib_mr_pool *mr_8k_pool;
+ unsigned int fmr_max_remaps;
+ unsigned int max_8k_fmrs;
+ unsigned int max_1m_fmrs;
int max_sge;
unsigned int max_wrs;
unsigned int max_initiator_depth;
struct rds_ib_statistics {
uint64_t s_ib_connect_raced;
uint64_t s_ib_listen_closed_stale;
- uint64_t s_ib_tx_cq_call;
+ uint64_t s_ib_evt_handler_call;
+ uint64_t s_ib_tasklet_call;
uint64_t s_ib_tx_cq_event;
uint64_t s_ib_tx_ring_full;
uint64_t s_ib_tx_throttle;
uint64_t s_ib_tx_sg_mapping_failure;
uint64_t s_ib_tx_stalled;
uint64_t s_ib_tx_credit_updates;
- uint64_t s_ib_rx_cq_call;
uint64_t s_ib_rx_cq_event;
uint64_t s_ib_rx_ring_empty;
uint64_t s_ib_rx_refill_from_cq;
uint64_t s_ib_ack_send_delayed;
uint64_t s_ib_ack_send_piggybacked;
uint64_t s_ib_ack_received;
- uint64_t s_ib_rdma_mr_alloc;
- uint64_t s_ib_rdma_mr_free;
- uint64_t s_ib_rdma_mr_used;
- uint64_t s_ib_rdma_mr_pool_flush;
- uint64_t s_ib_rdma_mr_pool_wait;
- uint64_t s_ib_rdma_mr_pool_depleted;
+ uint64_t s_ib_rdma_mr_8k_alloc;
+ uint64_t s_ib_rdma_mr_8k_free;
+ uint64_t s_ib_rdma_mr_8k_used;
+ uint64_t s_ib_rdma_mr_8k_pool_flush;
+ uint64_t s_ib_rdma_mr_8k_pool_wait;
+ uint64_t s_ib_rdma_mr_8k_pool_depleted;
+ uint64_t s_ib_rdma_mr_1m_alloc;
+ uint64_t s_ib_rdma_mr_1m_free;
+ uint64_t s_ib_rdma_mr_1m_used;
+ uint64_t s_ib_rdma_mr_1m_pool_flush;
+ uint64_t s_ib_rdma_mr_1m_pool_wait;
+ uint64_t s_ib_rdma_mr_1m_pool_depleted;
uint64_t s_ib_atomic_cswp;
uint64_t s_ib_atomic_fadd;
};
void rds_ib_dev_put(struct rds_ib_device *rds_ibdev);
extern struct ib_client rds_ib_client;
-extern unsigned int fmr_message_size;
+extern unsigned int rds_ib_fmr_1m_pool_size;
+extern unsigned int rds_ib_fmr_8k_pool_size;
extern unsigned int rds_ib_retry_count;
extern spinlock_t ib_nodev_conns_lock;
void rds_ib_add_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn);
void rds_ib_remove_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn);
void rds_ib_destroy_nodev_conns(void);
-struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *);
+struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_dev,
+ int npages);
void rds_ib_get_mr_info(struct rds_ib_device *rds_ibdev, struct rds_info_rdma_connection *iinfo);
void rds_ib_destroy_mr_pool(struct rds_ib_mr_pool *);
void *rds_ib_get_mr(struct scatterlist *sg, unsigned long nents,
void rds_ib_recv_refill(struct rds_connection *conn, int prefill, gfp_t gfp);
void rds_ib_inc_free(struct rds_incoming *inc);
int rds_ib_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to);
-void rds_ib_recv_cq_comp_handler(struct ib_cq *cq, void *context);
+void rds_ib_recv_cqe_handler(struct rds_ib_connection *ic, struct ib_wc *wc,
+ struct rds_ib_ack_state *state);
void rds_ib_recv_tasklet_fn(unsigned long data);
void rds_ib_recv_init_ring(struct rds_ib_connection *ic);
void rds_ib_recv_clear_ring(struct rds_ib_connection *ic);
void rds_ib_attempt_ack(struct rds_ib_connection *ic);
void rds_ib_ack_send_complete(struct rds_ib_connection *ic);
u64 rds_ib_piggyb_ack(struct rds_ib_connection *ic);
+void rds_ib_set_ack(struct rds_ib_connection *ic, u64 seq, int ack_required);
/* ib_ring.c */
void rds_ib_ring_init(struct rds_ib_work_ring *ring, u32 nr);
void rds_ib_xmit_complete(struct rds_connection *conn);
int rds_ib_xmit(struct rds_connection *conn, struct rds_message *rm,
unsigned int hdr_off, unsigned int sg, unsigned int off);
-void rds_ib_send_cq_comp_handler(struct ib_cq *cq, void *context);
+void rds_ib_send_cqe_handler(struct rds_ib_connection *ic, struct ib_wc *wc);
void rds_ib_send_init_ring(struct rds_ib_connection *ic);
void rds_ib_send_clear_ring(struct rds_ib_connection *ic);
int rds_ib_xmit_rdma(struct rds_connection *conn, struct rm_rdma_op *op);
event->event, ib_event_msg(event->event), data);
}
+/* Plucking the oldest entry from the ring can be done concurrently with
+ * the thread refilling the ring. Each ring operation is protected by
+ * spinlocks and the transient state of refilling doesn't change the
+ * recording of which entry is oldest.
+ *
+ * This relies on IB only calling one cq comp_handler for each cq so that
+ * there will only be one caller of rds_recv_incoming() per RDS connection.
+ */
+static void rds_ib_cq_comp_handler_recv(struct ib_cq *cq, void *context)
+{
+ struct rds_connection *conn = context;
+ struct rds_ib_connection *ic = conn->c_transport_data;
+
+ rdsdebug("conn %p cq %p\n", conn, cq);
+
+ rds_ib_stats_inc(s_ib_evt_handler_call);
+
+ tasklet_schedule(&ic->i_recv_tasklet);
+}
+
+static void poll_cq(struct rds_ib_connection *ic, struct ib_cq *cq,
+ struct ib_wc *wcs,
+ struct rds_ib_ack_state *ack_state)
+{
+ int nr;
+ int i;
+ struct ib_wc *wc;
+
+ while ((nr = ib_poll_cq(cq, RDS_IB_WC_MAX, wcs)) > 0) {
+ for (i = 0; i < nr; i++) {
+ wc = wcs + i;
+ rdsdebug("wc wr_id 0x%llx status %u byte_len %u imm_data %u\n",
+ (unsigned long long)wc->wr_id, wc->status,
+ wc->byte_len, be32_to_cpu(wc->ex.imm_data));
+
+ if (wc->wr_id & RDS_IB_SEND_OP)
+ rds_ib_send_cqe_handler(ic, wc);
+ else
+ rds_ib_recv_cqe_handler(ic, wc, ack_state);
+ }
+ }
+}
+
+static void rds_ib_tasklet_fn_send(unsigned long data)
+{
+ struct rds_ib_connection *ic = (struct rds_ib_connection *)data;
+ struct rds_connection *conn = ic->conn;
+ struct rds_ib_ack_state state;
+
+ rds_ib_stats_inc(s_ib_tasklet_call);
+
+ memset(&state, 0, sizeof(state));
+ poll_cq(ic, ic->i_send_cq, ic->i_send_wc, &state);
+ ib_req_notify_cq(ic->i_send_cq, IB_CQ_NEXT_COMP);
+ poll_cq(ic, ic->i_send_cq, ic->i_send_wc, &state);
+
+ if (rds_conn_up(conn) &&
+ (!test_bit(RDS_LL_SEND_FULL, &conn->c_flags) ||
+ test_bit(0, &conn->c_map_queued)))
+ rds_send_xmit(ic->conn);
+}
+
+static void rds_ib_tasklet_fn_recv(unsigned long data)
+{
+ struct rds_ib_connection *ic = (struct rds_ib_connection *)data;
+ struct rds_connection *conn = ic->conn;
+ struct rds_ib_device *rds_ibdev = ic->rds_ibdev;
+ struct rds_ib_ack_state state;
+
+ if (!rds_ibdev)
+ rds_conn_drop(conn);
+
+ rds_ib_stats_inc(s_ib_tasklet_call);
+
+ memset(&state, 0, sizeof(state));
+ poll_cq(ic, ic->i_recv_cq, ic->i_recv_wc, &state);
+ ib_req_notify_cq(ic->i_recv_cq, IB_CQ_SOLICITED);
+ poll_cq(ic, ic->i_recv_cq, ic->i_recv_wc, &state);
+
+ if (state.ack_next_valid)
+ rds_ib_set_ack(ic, state.ack_next, state.ack_required);
+ if (state.ack_recv_valid && state.ack_recv > ic->i_ack_recv) {
+ rds_send_drop_acked(conn, state.ack_recv, NULL);
+ ic->i_ack_recv = state.ack_recv;
+ }
+
+ if (rds_conn_up(conn))
+ rds_ib_attempt_ack(ic);
+}
+
static void rds_ib_qp_event_handler(struct ib_event *event, void *data)
{
struct rds_connection *conn = data;
}
}
+static void rds_ib_cq_comp_handler_send(struct ib_cq *cq, void *context)
+{
+ struct rds_connection *conn = context;
+ struct rds_ib_connection *ic = conn->c_transport_data;
+
+ rdsdebug("conn %p cq %p\n", conn, cq);
+
+ rds_ib_stats_inc(s_ib_evt_handler_call);
+
+ tasklet_schedule(&ic->i_send_tasklet);
+}
+
/*
* This needs to be very careful to not leave IS_ERR pointers around for
* cleanup to trip over.
ic->i_pd = rds_ibdev->pd;
cq_attr.cqe = ic->i_send_ring.w_nr + 1;
- ic->i_send_cq = ib_create_cq(dev, rds_ib_send_cq_comp_handler,
+
+ ic->i_send_cq = ib_create_cq(dev, rds_ib_cq_comp_handler_send,
rds_ib_cq_event_handler, conn,
&cq_attr);
if (IS_ERR(ic->i_send_cq)) {
}
cq_attr.cqe = ic->i_recv_ring.w_nr;
- ic->i_recv_cq = ib_create_cq(dev, rds_ib_recv_cq_comp_handler,
+ ic->i_recv_cq = ib_create_cq(dev, rds_ib_cq_comp_handler_recv,
rds_ib_cq_event_handler, conn,
&cq_attr);
if (IS_ERR(ic->i_recv_cq)) {
/* XXX I wonder what affect the port space has */
/* delegate cm event handler to rdma_transport */
- ic->i_cm_id = rdma_create_id(rds_rdma_cm_event_handler, conn,
+ ic->i_cm_id = rdma_create_id(&init_net, rds_rdma_cm_event_handler, conn,
RDMA_PS_TCP, IB_QPT_RC);
if (IS_ERR(ic->i_cm_id)) {
ret = PTR_ERR(ic->i_cm_id);
wait_event(rds_ib_ring_empty_wait,
rds_ib_ring_empty(&ic->i_recv_ring) &&
(atomic_read(&ic->i_signaled_sends) == 0));
+ tasklet_kill(&ic->i_send_tasklet);
tasklet_kill(&ic->i_recv_tasklet);
/* first destroy the ib state that generates callbacks */
}
INIT_LIST_HEAD(&ic->ib_node);
- tasklet_init(&ic->i_recv_tasklet, rds_ib_recv_tasklet_fn,
- (unsigned long) ic);
+ tasklet_init(&ic->i_send_tasklet, rds_ib_tasklet_fn_send,
+ (unsigned long)ic);
+ tasklet_init(&ic->i_recv_tasklet, rds_ib_tasklet_fn_recv,
+ (unsigned long)ic);
mutex_init(&ic->i_recv_mutex);
#ifndef KERNEL_HAS_ATOMIC64
spin_lock_init(&ic->i_ack_lock);
send->s_op = NULL;
- send->s_wr.wr_id = i;
+ send->s_wr.wr_id = i | RDS_IB_SEND_OP;
send->s_wr.sg_list = send->s_sge;
send->s_wr.ex.imm_data = 0;
* unallocs the next free entry in the ring it doesn't alter which is
* the next to be freed, which is what this is concerned with.
*/
-void rds_ib_send_cq_comp_handler(struct ib_cq *cq, void *context)
+void rds_ib_send_cqe_handler(struct rds_ib_connection *ic, struct ib_wc *wc)
{
- struct rds_connection *conn = context;
- struct rds_ib_connection *ic = conn->c_transport_data;
struct rds_message *rm = NULL;
- struct ib_wc wc;
+ struct rds_connection *conn = ic->conn;
struct rds_ib_send_work *send;
u32 completed;
u32 oldest;
u32 i = 0;
- int ret;
int nr_sig = 0;
- rdsdebug("cq %p conn %p\n", cq, conn);
- rds_ib_stats_inc(s_ib_tx_cq_call);
- ret = ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
- if (ret)
- rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
-
- while (ib_poll_cq(cq, 1, &wc) > 0) {
- rdsdebug("wc wr_id 0x%llx status %u (%s) byte_len %u imm_data %u\n",
- (unsigned long long)wc.wr_id, wc.status,
- ib_wc_status_msg(wc.status), wc.byte_len,
- be32_to_cpu(wc.ex.imm_data));
- rds_ib_stats_inc(s_ib_tx_cq_event);
-
- if (wc.wr_id == RDS_IB_ACK_WR_ID) {
- if (time_after(jiffies, ic->i_ack_queued + HZ/2))
- rds_ib_stats_inc(s_ib_tx_stalled);
- rds_ib_ack_send_complete(ic);
- continue;
- }
- oldest = rds_ib_ring_oldest(&ic->i_send_ring);
+ rdsdebug("wc wr_id 0x%llx status %u (%s) byte_len %u imm_data %u\n",
+ (unsigned long long)wc->wr_id, wc->status,
+ ib_wc_status_msg(wc->status), wc->byte_len,
+ be32_to_cpu(wc->ex.imm_data));
+ rds_ib_stats_inc(s_ib_tx_cq_event);
- completed = rds_ib_ring_completed(&ic->i_send_ring, wc.wr_id, oldest);
+ if (wc->wr_id == RDS_IB_ACK_WR_ID) {
+ if (time_after(jiffies, ic->i_ack_queued + HZ / 2))
+ rds_ib_stats_inc(s_ib_tx_stalled);
+ rds_ib_ack_send_complete(ic);
+ return;
+ }
- for (i = 0; i < completed; i++) {
- send = &ic->i_sends[oldest];
- if (send->s_wr.send_flags & IB_SEND_SIGNALED)
- nr_sig++;
+ oldest = rds_ib_ring_oldest(&ic->i_send_ring);
- rm = rds_ib_send_unmap_op(ic, send, wc.status);
+ completed = rds_ib_ring_completed(&ic->i_send_ring,
+ (wc->wr_id & ~RDS_IB_SEND_OP),
+ oldest);
- if (time_after(jiffies, send->s_queued + HZ/2))
- rds_ib_stats_inc(s_ib_tx_stalled);
+ for (i = 0; i < completed; i++) {
+ send = &ic->i_sends[oldest];
+ if (send->s_wr.send_flags & IB_SEND_SIGNALED)
+ nr_sig++;
- if (send->s_op) {
- if (send->s_op == rm->m_final_op) {
- /* If anyone waited for this message to get flushed out, wake
- * them up now */
- rds_message_unmapped(rm);
- }
- rds_message_put(rm);
- send->s_op = NULL;
- }
+ rm = rds_ib_send_unmap_op(ic, send, wc->status);
+
+ if (time_after(jiffies, send->s_queued + HZ / 2))
+ rds_ib_stats_inc(s_ib_tx_stalled);
- oldest = (oldest + 1) % ic->i_send_ring.w_nr;
+ if (send->s_op) {
+ if (send->s_op == rm->m_final_op) {
+ /* If anyone waited for this message to get
+ * flushed out, wake them up now
+ */
+ rds_message_unmapped(rm);
+ }
+ rds_message_put(rm);
+ send->s_op = NULL;
}
- rds_ib_ring_free(&ic->i_send_ring, completed);
- rds_ib_sub_signaled(ic, nr_sig);
- nr_sig = 0;
+ oldest = (oldest + 1) % ic->i_send_ring.w_nr;
+ }
- if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags) ||
- test_bit(0, &conn->c_map_queued))
- queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+ rds_ib_ring_free(&ic->i_send_ring, completed);
+ rds_ib_sub_signaled(ic, nr_sig);
+ nr_sig = 0;
- /* We expect errors as the qp is drained during shutdown */
- if (wc.status != IB_WC_SUCCESS && rds_conn_up(conn)) {
- rds_ib_conn_error(conn, "send completion on %pI4 had status "
- "%u (%s), disconnecting and reconnecting\n",
- &conn->c_faddr, wc.status,
- ib_wc_status_msg(wc.status));
- }
+ if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags) ||
+ test_bit(0, &conn->c_map_queued))
+ queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+
+ /* We expect errors as the qp is drained during shutdown */
+ if (wc->status != IB_WC_SUCCESS && rds_conn_up(conn)) {
+ rds_ib_conn_error(conn, "send completion on %pI4 had status %u (%s), disconnecting and reconnecting\n",
+ &conn->c_faddr, wc->status,
+ ib_wc_status_msg(wc->status));
}
}
send->s_queued = jiffies;
if (op->op_type == RDS_ATOMIC_TYPE_CSWP) {
- send->s_wr.opcode = IB_WR_MASKED_ATOMIC_CMP_AND_SWP;
- send->s_wr.wr.atomic.compare_add = op->op_m_cswp.compare;
- send->s_wr.wr.atomic.swap = op->op_m_cswp.swap;
- send->s_wr.wr.atomic.compare_add_mask = op->op_m_cswp.compare_mask;
- send->s_wr.wr.atomic.swap_mask = op->op_m_cswp.swap_mask;
+ send->s_atomic_wr.wr.opcode = IB_WR_MASKED_ATOMIC_CMP_AND_SWP;
+ send->s_atomic_wr.compare_add = op->op_m_cswp.compare;
+ send->s_atomic_wr.swap = op->op_m_cswp.swap;
+ send->s_atomic_wr.compare_add_mask = op->op_m_cswp.compare_mask;
+ send->s_atomic_wr.swap_mask = op->op_m_cswp.swap_mask;
} else { /* FADD */
- send->s_wr.opcode = IB_WR_MASKED_ATOMIC_FETCH_AND_ADD;
- send->s_wr.wr.atomic.compare_add = op->op_m_fadd.add;
- send->s_wr.wr.atomic.swap = 0;
- send->s_wr.wr.atomic.compare_add_mask = op->op_m_fadd.nocarry_mask;
- send->s_wr.wr.atomic.swap_mask = 0;
+ send->s_atomic_wr.wr.opcode = IB_WR_MASKED_ATOMIC_FETCH_AND_ADD;
+ send->s_atomic_wr.compare_add = op->op_m_fadd.add;
+ send->s_atomic_wr.swap = 0;
+ send->s_atomic_wr.compare_add_mask = op->op_m_fadd.nocarry_mask;
+ send->s_atomic_wr.swap_mask = 0;
}
nr_sig = rds_ib_set_wr_signal_state(ic, send, op->op_notify);
- send->s_wr.num_sge = 1;
- send->s_wr.next = NULL;
- send->s_wr.wr.atomic.remote_addr = op->op_remote_addr;
- send->s_wr.wr.atomic.rkey = op->op_rkey;
+ send->s_atomic_wr.wr.num_sge = 1;
+ send->s_atomic_wr.wr.next = NULL;
+ send->s_atomic_wr.remote_addr = op->op_remote_addr;
+ send->s_atomic_wr.rkey = op->op_rkey;
send->s_op = op;
rds_message_addref(container_of(send->s_op, struct rds_message, atomic));
if (nr_sig)
atomic_add(nr_sig, &ic->i_signaled_sends);
- failed_wr = &send->s_wr;
- ret = ib_post_send(ic->i_cm_id->qp, &send->s_wr, &failed_wr);
+ failed_wr = &send->s_atomic_wr.wr;
+ ret = ib_post_send(ic->i_cm_id->qp, &send->s_atomic_wr.wr, &failed_wr);
rdsdebug("ic %p send %p (wr %p) ret %d wr %p\n", ic,
- send, &send->s_wr, ret, failed_wr);
- BUG_ON(failed_wr != &send->s_wr);
+ send, &send->s_atomic_wr, ret, failed_wr);
+ BUG_ON(failed_wr != &send->s_atomic_wr.wr);
if (ret) {
printk(KERN_WARNING "RDS/IB: atomic ib_post_send to %pI4 "
"returned %d\n", &conn->c_faddr, ret);
goto out;
}
- if (unlikely(failed_wr != &send->s_wr)) {
+ if (unlikely(failed_wr != &send->s_atomic_wr.wr)) {
printk(KERN_WARNING "RDS/IB: atomic ib_post_send() rc=%d, but failed_wqe updated!\n", ret);
- BUG_ON(failed_wr != &send->s_wr);
+ BUG_ON(failed_wr != &send->s_atomic_wr.wr);
}
out:
nr_sig += rds_ib_set_wr_signal_state(ic, send, op->op_notify);
send->s_wr.opcode = op->op_write ? IB_WR_RDMA_WRITE : IB_WR_RDMA_READ;
- send->s_wr.wr.rdma.remote_addr = remote_addr;
- send->s_wr.wr.rdma.rkey = op->op_rkey;
+ send->s_rdma_wr.remote_addr = remote_addr;
+ send->s_rdma_wr.rkey = op->op_rkey;
if (num_sge > max_sge) {
- send->s_wr.num_sge = max_sge;
+ send->s_rdma_wr.wr.num_sge = max_sge;
num_sge -= max_sge;
} else {
- send->s_wr.num_sge = num_sge;
+ send->s_rdma_wr.wr.num_sge = num_sge;
}
- send->s_wr.next = NULL;
+ send->s_rdma_wr.wr.next = NULL;
if (prev)
- prev->s_wr.next = &send->s_wr;
+ prev->s_rdma_wr.wr.next = &send->s_rdma_wr.wr;
- for (j = 0; j < send->s_wr.num_sge && scat != &op->op_sg[op->op_count]; j++) {
+ for (j = 0; j < send->s_rdma_wr.wr.num_sge &&
+ scat != &op->op_sg[op->op_count]; j++) {
len = ib_sg_dma_len(ic->i_cm_id->device, scat);
send->s_sge[j].addr =
ib_sg_dma_address(ic->i_cm_id->device, scat);
}
rdsdebug("send %p wr %p num_sge %u next %p\n", send,
- &send->s_wr, send->s_wr.num_sge, send->s_wr.next);
+ &send->s_rdma_wr.wr,
+ send->s_rdma_wr.wr.num_sge,
+ send->s_rdma_wr.wr.next);
prev = send;
if (++send == &ic->i_sends[ic->i_send_ring.w_nr])
if (nr_sig)
atomic_add(nr_sig, &ic->i_signaled_sends);
- failed_wr = &first->s_wr;
- ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
+ failed_wr = &first->s_rdma_wr.wr;
+ ret = ib_post_send(ic->i_cm_id->qp, &first->s_rdma_wr.wr, &failed_wr);
rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
- first, &first->s_wr, ret, failed_wr);
- BUG_ON(failed_wr != &first->s_wr);
+ first, &first->s_rdma_wr.wr, ret, failed_wr);
+ BUG_ON(failed_wr != &first->s_rdma_wr.wr);
if (ret) {
printk(KERN_WARNING "RDS/IB: rdma ib_post_send to %pI4 "
"returned %d\n", &conn->c_faddr, ret);
goto out;
}
- if (unlikely(failed_wr != &first->s_wr)) {
+ if (unlikely(failed_wr != &first->s_rdma_wr.wr)) {
printk(KERN_WARNING "RDS/IB: ib_post_send() rc=%d, but failed_wqe updated!\n", ret);
- BUG_ON(failed_wr != &first->s_wr);
+ BUG_ON(failed_wr != &first->s_rdma_wr.wr);
}
struct rdma_cm_id *cm_id;
struct ib_mr *mr;
- struct ib_fast_reg_page_list *page_list;
struct rds_iw_mapping mapping;
unsigned char remap_count;
int max_pages;
};
-static int rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all);
+static void rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all);
static void rds_iw_mr_pool_flush_worker(struct work_struct *work);
- static int rds_iw_init_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
- static int rds_iw_map_fastreg(struct rds_iw_mr_pool *pool,
+ static int rds_iw_init_reg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
+ static int rds_iw_map_reg(struct rds_iw_mr_pool *pool,
struct rds_iw_mr *ibmr,
struct scatterlist *sg, unsigned int nents);
static void rds_iw_free_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
sg->bytes = 0;
}
- static u64 *rds_iw_map_scatterlist(struct rds_iw_device *rds_iwdev,
- struct rds_iw_scatterlist *sg)
+ static int rds_iw_map_scatterlist(struct rds_iw_device *rds_iwdev,
+ struct rds_iw_scatterlist *sg)
{
struct ib_device *dev = rds_iwdev->dev;
- u64 *dma_pages = NULL;
- int i, j, ret;
+ int i, ret;
WARN_ON(sg->dma_len);
sg->dma_len = ib_dma_map_sg(dev, sg->list, sg->len, DMA_BIDIRECTIONAL);
if (unlikely(!sg->dma_len)) {
printk(KERN_WARNING "RDS/IW: dma_map_sg failed!\n");
- return ERR_PTR(-EBUSY);
+ return -EBUSY;
}
sg->bytes = 0;
if (sg->dma_npages > fastreg_message_size)
goto out_unmap;
- dma_pages = kmalloc(sizeof(u64) * sg->dma_npages, GFP_ATOMIC);
- if (!dma_pages) {
- ret = -ENOMEM;
- goto out_unmap;
- }
- for (i = j = 0; i < sg->dma_len; ++i) {
- unsigned int dma_len = ib_sg_dma_len(dev, &sg->list[i]);
- u64 dma_addr = ib_sg_dma_address(dev, &sg->list[i]);
- u64 end_addr;
- end_addr = dma_addr + dma_len;
- dma_addr &= ~PAGE_MASK;
- for (; dma_addr < end_addr; dma_addr += PAGE_SIZE)
- dma_pages[j++] = dma_addr;
- BUG_ON(j > sg->dma_npages);
- }
-
- return dma_pages;
+ return 0;
out_unmap:
ib_dma_unmap_sg(rds_iwdev->dev, sg->list, sg->len, DMA_BIDIRECTIONAL);
sg->dma_len = 0;
- kfree(dma_pages);
- return ERR_PTR(ret);
+ return ret;
}
INIT_LIST_HEAD(&ibmr->mapping.m_list);
ibmr->mapping.m_mr = ibmr;
- err = rds_iw_init_fastreg(pool, ibmr);
+ err = rds_iw_init_reg(pool, ibmr);
if (err)
goto out_no_cigar;
* If the number of MRs allocated exceeds the limit, we also try
* to free as many MRs as needed to get back to this limit.
*/
-static int rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all)
+static void rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all)
{
struct rds_iw_mr *ibmr, *next;
LIST_HEAD(unmap_list);
LIST_HEAD(kill_list);
unsigned long flags;
unsigned int nfreed = 0, ncleaned = 0, unpinned = 0;
- int ret = 0;
rds_iw_stats_inc(s_iw_rdma_mr_pool_flush);
atomic_sub(nfreed, &pool->item_count);
mutex_unlock(&pool->flush_lock);
- return ret;
}
static void rds_iw_mr_pool_flush_worker(struct work_struct *work)
ibmr->cm_id = cm_id;
ibmr->device = rds_iwdev;
- ret = rds_iw_map_fastreg(rds_iwdev->mr_pool, ibmr, sg, nents);
+ ret = rds_iw_map_reg(rds_iwdev->mr_pool, ibmr, sg, nents);
if (ret == 0)
*key_ret = ibmr->mr->rkey;
else
}
/*
- * iWARP fastreg handling
+ * iWARP reg handling
*
* The life cycle of a fastreg registration is a bit different from
* FMRs.
* This creates a bit of a problem for us, as we do not have the destination
* IP in GET_MR, so the connection must be setup prior to the GET_MR call for
* RDMA to be correctly setup. If a fastreg request is present, rds_iw_xmit
- * will try to queue a LOCAL_INV (if needed) and a FAST_REG_MR work request
+ * will try to queue a LOCAL_INV (if needed) and a REG_MR work request
* before queuing the SEND. When completions for these arrive, they are
* dispatched to the MR has a bit set showing that RDMa can be performed.
*
* The expectation there is that this invalidation step includes ALL
* PREVIOUSLY FREED MRs.
*/
- static int rds_iw_init_fastreg(struct rds_iw_mr_pool *pool,
- struct rds_iw_mr *ibmr)
+ static int rds_iw_init_reg(struct rds_iw_mr_pool *pool,
+ struct rds_iw_mr *ibmr)
{
struct rds_iw_device *rds_iwdev = pool->device;
- struct ib_fast_reg_page_list *page_list = NULL;
struct ib_mr *mr;
int err;
return err;
}
- /* FIXME - this is overkill, but mapping->m_sg.dma_len/mapping->m_sg.dma_npages
- * is not filled in.
- */
- page_list = ib_alloc_fast_reg_page_list(rds_iwdev->dev, pool->max_message_size);
- if (IS_ERR(page_list)) {
- err = PTR_ERR(page_list);
-
- printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_page_list failed (err=%d)\n", err);
- ib_dereg_mr(mr);
- return err;
- }
-
- ibmr->page_list = page_list;
ibmr->mr = mr;
return 0;
}
- static int rds_iw_rdma_build_fastreg(struct rds_iw_mapping *mapping)
+ static int rds_iw_rdma_reg_mr(struct rds_iw_mapping *mapping)
{
struct rds_iw_mr *ibmr = mapping->m_mr;
- struct ib_send_wr f_wr, *failed_wr;
- int ret;
+ struct rds_iw_scatterlist *m_sg = &mapping->m_sg;
+ struct ib_reg_wr reg_wr;
+ struct ib_send_wr *failed_wr;
+ int ret, n;
+
+ n = ib_map_mr_sg_zbva(ibmr->mr, m_sg->list, m_sg->len, PAGE_SIZE);
+ if (unlikely(n != m_sg->len))
+ return n < 0 ? n : -EINVAL;
+
+ reg_wr.wr.next = NULL;
+ reg_wr.wr.opcode = IB_WR_REG_MR;
+ reg_wr.wr.wr_id = RDS_IW_REG_WR_ID;
+ reg_wr.wr.num_sge = 0;
+ reg_wr.mr = ibmr->mr;
+ reg_wr.key = mapping->m_rkey;
+ reg_wr.access = IB_ACCESS_LOCAL_WRITE |
+ IB_ACCESS_REMOTE_READ |
+ IB_ACCESS_REMOTE_WRITE;
/*
- * Perform a WR for the fast_reg_mr. Each individual page
+ * Perform a WR for the reg_mr. Each individual page
* in the sg list is added to the fast reg page list and placed
- * inside the fast_reg_mr WR. The key used is a rolling 8bit
+ * inside the reg_mr WR. The key used is a rolling 8bit
* counter, which should guarantee uniqueness.
*/
ib_update_fast_reg_key(ibmr->mr, ibmr->remap_count++);
mapping->m_rkey = ibmr->mr->rkey;
- memset(&f_wr, 0, sizeof(f_wr));
- f_wr.wr_id = RDS_IW_FAST_REG_WR_ID;
- f_wr.opcode = IB_WR_FAST_REG_MR;
- f_wr.wr.fast_reg.length = mapping->m_sg.bytes;
- f_wr.wr.fast_reg.rkey = mapping->m_rkey;
- f_wr.wr.fast_reg.page_list = ibmr->page_list;
- f_wr.wr.fast_reg.page_list_len = mapping->m_sg.dma_len;
- f_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
- f_wr.wr.fast_reg.access_flags = IB_ACCESS_LOCAL_WRITE |
- IB_ACCESS_REMOTE_READ |
- IB_ACCESS_REMOTE_WRITE;
- f_wr.wr.fast_reg.iova_start = 0;
- f_wr.send_flags = IB_SEND_SIGNALED;
-
- failed_wr = &f_wr;
- ret = ib_post_send(ibmr->cm_id->qp, &f_wr, &failed_wr);
- BUG_ON(failed_wr != &f_wr);
+ failed_wr = ®_wr.wr;
+ ret = ib_post_send(ibmr->cm_id->qp, ®_wr.wr, &failed_wr);
+ BUG_ON(failed_wr != ®_wr.wr);
if (ret)
printk_ratelimited(KERN_WARNING "RDS/IW: %s:%d ib_post_send returned %d\n",
__func__, __LINE__, ret);
return ret;
}
- static int rds_iw_map_fastreg(struct rds_iw_mr_pool *pool,
- struct rds_iw_mr *ibmr,
- struct scatterlist *sg,
- unsigned int sg_len)
+ static int rds_iw_map_reg(struct rds_iw_mr_pool *pool,
+ struct rds_iw_mr *ibmr,
+ struct scatterlist *sg,
+ unsigned int sg_len)
{
struct rds_iw_device *rds_iwdev = pool->device;
struct rds_iw_mapping *mapping = &ibmr->mapping;
u64 *dma_pages;
- int i, ret = 0;
+ int ret = 0;
rds_iw_set_scatterlist(&mapping->m_sg, sg, sg_len);
- dma_pages = rds_iw_map_scatterlist(rds_iwdev, &mapping->m_sg);
- if (IS_ERR(dma_pages)) {
- ret = PTR_ERR(dma_pages);
+ ret = rds_iw_map_scatterlist(rds_iwdev, &mapping->m_sg);
+ if (ret) {
dma_pages = NULL;
goto out;
}
goto out;
}
- for (i = 0; i < mapping->m_sg.dma_npages; ++i)
- ibmr->page_list->page_list[i] = dma_pages[i];
-
- ret = rds_iw_rdma_build_fastreg(mapping);
+ ret = rds_iw_rdma_reg_mr(mapping);
if (ret)
goto out;
static void rds_iw_destroy_fastreg(struct rds_iw_mr_pool *pool,
struct rds_iw_mr *ibmr)
{
- if (ibmr->page_list)
- ib_free_fast_reg_page_list(ibmr->page_list);
if (ibmr->mr)
ib_dereg_mr(ibmr->mr);
}
f->fr_mr = ib_alloc_mr(pd, IB_MR_TYPE_MEM_REG, depth);
if (IS_ERR(f->fr_mr))
goto out_mr_err;
- f->fr_pgl = ib_alloc_fast_reg_page_list(device, depth);
- if (IS_ERR(f->fr_pgl))
+
+ f->sg = kcalloc(depth, sizeof(*f->sg), GFP_KERNEL);
+ if (!f->sg)
goto out_list_err;
+
+ sg_init_table(f->sg, depth);
+
return 0;
out_mr_err:
return rc;
out_list_err:
- rc = PTR_ERR(f->fr_pgl);
- dprintk("RPC: %s: ib_alloc_fast_reg_page_list status %i\n",
- __func__, rc);
+ rc = -ENOMEM;
+ dprintk("RPC: %s: sg allocation failure\n",
+ __func__);
ib_dereg_mr(f->fr_mr);
return rc;
}
if (rc)
dprintk("RPC: %s: ib_dereg_mr status %i\n",
__func__, rc);
- ib_free_fast_reg_page_list(r->r.frmr.fr_pgl);
+ kfree(r->r.frmr.sg);
}
static int
/* WARNING: Only wr_id and status are reliable at this point */
r = (struct rpcrdma_mw *)(unsigned long)wc->wr_id;
- pr_warn("RPC: %s: frmr %p flushed, status %s (%d)\n",
- __func__, r, ib_wc_status_msg(wc->status), wc->status);
+ if (wc->status == IB_WC_WR_FLUSH_ERR)
+ dprintk("RPC: %s: frmr %p flushed\n", __func__, r);
+ else
+ pr_warn("RPC: %s: frmr %p error, status %s (%d)\n",
+ __func__, r, ib_wc_status_msg(wc->status), wc->status);
r->r.frmr.fr_state = FRMR_IS_STALE;
}
struct rpcrdma_mw *mw;
struct rpcrdma_frmr *frmr;
struct ib_mr *mr;
- struct ib_send_wr fastreg_wr, *bad_wr;
+ struct ib_reg_wr reg_wr;
+ struct ib_send_wr *bad_wr;
+ int rc, i, n, dma_nents;
u8 key;
- int len, pageoff;
- int i, rc;
- int seg_len;
- u64 pa;
- int page_no;
mw = seg1->rl_mw;
seg1->rl_mw = NULL;
} while (mw->r.frmr.fr_state != FRMR_IS_INVALID);
frmr = &mw->r.frmr;
frmr->fr_state = FRMR_IS_VALID;
+ mr = frmr->fr_mr;
- pageoff = offset_in_page(seg1->mr_offset);
- seg1->mr_offset -= pageoff; /* start of page */
- seg1->mr_len += pageoff;
- len = -pageoff;
if (nsegs > ia->ri_max_frmr_depth)
nsegs = ia->ri_max_frmr_depth;
- for (page_no = i = 0; i < nsegs;) {
- rpcrdma_map_one(device, seg, direction);
- pa = seg->mr_dma;
- for (seg_len = seg->mr_len; seg_len > 0; seg_len -= PAGE_SIZE) {
- frmr->fr_pgl->page_list[page_no++] = pa;
- pa += PAGE_SIZE;
- }
- len += seg->mr_len;
+ for (i = 0; i < nsegs;) {
+ if (seg->mr_page)
+ sg_set_page(&frmr->sg[i],
+ seg->mr_page,
+ seg->mr_len,
+ offset_in_page(seg->mr_offset));
+ else
+ sg_set_buf(&frmr->sg[i], seg->mr_offset,
+ seg->mr_len);
+
++seg;
++i;
+
/* Check for holes */
if ((i < nsegs && offset_in_page(seg->mr_offset)) ||
offset_in_page((seg-1)->mr_offset + (seg-1)->mr_len))
break;
}
- dprintk("RPC: %s: Using frmr %p to map %d segments (%d bytes)\n",
- __func__, mw, i, len);
-
- memset(&fastreg_wr, 0, sizeof(fastreg_wr));
- fastreg_wr.wr_id = (unsigned long)(void *)mw;
- fastreg_wr.opcode = IB_WR_FAST_REG_MR;
- fastreg_wr.wr.fast_reg.iova_start = seg1->mr_dma + pageoff;
- fastreg_wr.wr.fast_reg.page_list = frmr->fr_pgl;
- fastreg_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
- fastreg_wr.wr.fast_reg.page_list_len = page_no;
- fastreg_wr.wr.fast_reg.length = len;
- fastreg_wr.wr.fast_reg.access_flags = writing ?
- IB_ACCESS_REMOTE_WRITE | IB_ACCESS_LOCAL_WRITE :
- IB_ACCESS_REMOTE_READ;
- mr = frmr->fr_mr;
+ frmr->sg_nents = i;
+
+ dma_nents = ib_dma_map_sg(device, frmr->sg, frmr->sg_nents, direction);
+ if (!dma_nents) {
+ pr_err("RPC: %s: failed to dma map sg %p sg_nents %u\n",
+ __func__, frmr->sg, frmr->sg_nents);
+ return -ENOMEM;
+ }
+
+ n = ib_map_mr_sg(mr, frmr->sg, frmr->sg_nents, PAGE_SIZE);
+ if (unlikely(n != frmr->sg_nents)) {
+ pr_err("RPC: %s: failed to map mr %p (%u/%u)\n",
+ __func__, frmr->fr_mr, n, frmr->sg_nents);
+ rc = n < 0 ? n : -EINVAL;
+ goto out_senderr;
+ }
+
+ dprintk("RPC: %s: Using frmr %p to map %u segments (%u bytes)\n",
+ __func__, mw, frmr->sg_nents, mr->length);
+
key = (u8)(mr->rkey & 0x000000FF);
ib_update_fast_reg_key(mr, ++key);
- fastreg_wr.wr.fast_reg.rkey = mr->rkey;
+
+ reg_wr.wr.next = NULL;
+ reg_wr.wr.opcode = IB_WR_REG_MR;
+ reg_wr.wr.wr_id = (uintptr_t)mw;
+ reg_wr.wr.num_sge = 0;
+ reg_wr.wr.send_flags = 0;
+ reg_wr.mr = mr;
+ reg_wr.key = mr->rkey;
+ reg_wr.access = writing ?
+ IB_ACCESS_REMOTE_WRITE | IB_ACCESS_LOCAL_WRITE :
+ IB_ACCESS_REMOTE_READ;
DECR_CQCOUNT(&r_xprt->rx_ep);
- rc = ib_post_send(ia->ri_id->qp, &fastreg_wr, &bad_wr);
+ rc = ib_post_send(ia->ri_id->qp, ®_wr.wr, &bad_wr);
if (rc)
goto out_senderr;
+ seg1->mr_dir = direction;
seg1->rl_mw = mw;
seg1->mr_rkey = mr->rkey;
- seg1->mr_base = seg1->mr_dma + pageoff;
- seg1->mr_nsegs = i;
- seg1->mr_len = len;
- return i;
+ seg1->mr_base = mr->iova;
+ seg1->mr_nsegs = frmr->sg_nents;
+ seg1->mr_len = mr->length;
+
+ return frmr->sg_nents;
out_senderr:
dprintk("RPC: %s: ib_post_send status %i\n", __func__, rc);
- while (i--)
- rpcrdma_unmap_one(device, --seg);
+ ib_dma_unmap_sg(device, frmr->sg, dma_nents, direction);
__frwr_queue_recovery(mw);
return rc;
}
struct rpcrdma_mr_seg *seg1 = seg;
struct rpcrdma_ia *ia = &r_xprt->rx_ia;
struct rpcrdma_mw *mw = seg1->rl_mw;
+ struct rpcrdma_frmr *frmr = &mw->r.frmr;
struct ib_send_wr invalidate_wr, *bad_wr;
int rc, nsegs = seg->mr_nsegs;
dprintk("RPC: %s: FRMR %p\n", __func__, mw);
seg1->rl_mw = NULL;
- mw->r.frmr.fr_state = FRMR_IS_INVALID;
+ frmr->fr_state = FRMR_IS_INVALID;
memset(&invalidate_wr, 0, sizeof(invalidate_wr));
invalidate_wr.wr_id = (unsigned long)(void *)mw;
invalidate_wr.opcode = IB_WR_LOCAL_INV;
- invalidate_wr.ex.invalidate_rkey = mw->r.frmr.fr_mr->rkey;
+ invalidate_wr.ex.invalidate_rkey = frmr->fr_mr->rkey;
DECR_CQCOUNT(&r_xprt->rx_ep);
- while (seg1->mr_nsegs--)
- rpcrdma_unmap_one(ia->ri_device, seg++);
+ ib_dma_unmap_sg(ia->ri_device, frmr->sg, frmr->sg_nents, seg1->mr_dir);
read_lock(&ia->ri_qplock);
rc = ib_post_send(ia->ri_id->qp, &invalidate_wr, &bad_wr);
read_unlock(&ia->ri_qplock);
u64 rs_offset,
bool last)
{
- struct ib_send_wr read_wr;
+ struct ib_rdma_wr read_wr;
int pages_needed = PAGE_ALIGN(*page_offset + rs_length) >> PAGE_SHIFT;
struct svc_rdma_op_ctxt *ctxt = svc_rdma_get_context(xprt);
int ret, read, pno;
ctxt->direction = DMA_FROM_DEVICE;
ctxt->read_hdr = head;
pages_needed = min_t(int, pages_needed, xprt->sc_max_sge_rd);
- read = min_t(int, pages_needed << PAGE_SHIFT, rs_length);
+ read = min_t(int, (pages_needed << PAGE_SHIFT) - *page_offset,
+ rs_length);
for (pno = 0; pno < pages_needed; pno++) {
int len = min_t(int, rs_length, PAGE_SIZE - pg_off);
clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
memset(&read_wr, 0, sizeof(read_wr));
- read_wr.wr_id = (unsigned long)ctxt;
- read_wr.opcode = IB_WR_RDMA_READ;
- ctxt->wr_op = read_wr.opcode;
- read_wr.send_flags = IB_SEND_SIGNALED;
- read_wr.wr.rdma.rkey = rs_handle;
- read_wr.wr.rdma.remote_addr = rs_offset;
- read_wr.sg_list = ctxt->sge;
- read_wr.num_sge = pages_needed;
-
- ret = svc_rdma_send(xprt, &read_wr);
+ read_wr.wr.wr_id = (unsigned long)ctxt;
+ read_wr.wr.opcode = IB_WR_RDMA_READ;
+ ctxt->wr_op = read_wr.wr.opcode;
+ read_wr.wr.send_flags = IB_SEND_SIGNALED;
+ read_wr.rkey = rs_handle;
+ read_wr.remote_addr = rs_offset;
+ read_wr.wr.sg_list = ctxt->sge;
+ read_wr.wr.num_sge = pages_needed;
+
+ ret = svc_rdma_send(xprt, &read_wr.wr);
if (ret) {
pr_err("svcrdma: Error %d posting RDMA_READ\n", ret);
set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags);
u64 rs_offset,
bool last)
{
- struct ib_send_wr read_wr;
+ struct ib_rdma_wr read_wr;
struct ib_send_wr inv_wr;
- struct ib_send_wr fastreg_wr;
+ struct ib_reg_wr reg_wr;
u8 key;
- int pages_needed = PAGE_ALIGN(*page_offset + rs_length) >> PAGE_SHIFT;
+ int nents = PAGE_ALIGN(*page_offset + rs_length) >> PAGE_SHIFT;
struct svc_rdma_op_ctxt *ctxt = svc_rdma_get_context(xprt);
struct svc_rdma_fastreg_mr *frmr = svc_rdma_get_frmr(xprt);
- int ret, read, pno;
+ int ret, read, pno, dma_nents, n;
u32 pg_off = *page_offset;
u32 pg_no = *page_no;
ctxt->direction = DMA_FROM_DEVICE;
ctxt->frmr = frmr;
- pages_needed = min_t(int, pages_needed, xprt->sc_frmr_pg_list_len);
- read = min_t(int, (pages_needed << PAGE_SHIFT) - *page_offset,
- rs_length);
+ nents = min_t(unsigned int, nents, xprt->sc_frmr_pg_list_len);
- read = min_t(int, nents << PAGE_SHIFT, rs_length);
++ read = min_t(int, (nents << PAGE_SHIFT) - *page_offset, rs_length);
- frmr->kva = page_address(rqstp->rq_arg.pages[pg_no]);
frmr->direction = DMA_FROM_DEVICE;
frmr->access_flags = (IB_ACCESS_LOCAL_WRITE|IB_ACCESS_REMOTE_WRITE);
- frmr->map_len = pages_needed << PAGE_SHIFT;
- frmr->page_list_len = pages_needed;
+ frmr->sg_nents = nents;
- for (pno = 0; pno < pages_needed; pno++) {
+ for (pno = 0; pno < nents; pno++) {
int len = min_t(int, rs_length, PAGE_SIZE - pg_off);
head->arg.pages[pg_no] = rqstp->rq_arg.pages[pg_no];
head->arg.len += len;
if (!pg_off)
head->count++;
+
+ sg_set_page(&frmr->sg[pno], rqstp->rq_arg.pages[pg_no],
+ len, pg_off);
+
rqstp->rq_respages = &rqstp->rq_arg.pages[pg_no+1];
rqstp->rq_next_page = rqstp->rq_respages + 1;
- frmr->page_list->page_list[pno] =
- ib_dma_map_page(xprt->sc_cm_id->device,
- head->arg.pages[pg_no], 0,
- PAGE_SIZE, DMA_FROM_DEVICE);
- ret = ib_dma_mapping_error(xprt->sc_cm_id->device,
- frmr->page_list->page_list[pno]);
- if (ret)
- goto err;
- atomic_inc(&xprt->sc_dma_used);
/* adjust offset and wrap to next page if needed */
pg_off += len;
else
clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
+ dma_nents = ib_dma_map_sg(xprt->sc_cm_id->device,
+ frmr->sg, frmr->sg_nents,
+ frmr->direction);
+ if (!dma_nents) {
+ pr_err("svcrdma: failed to dma map sg %p\n",
+ frmr->sg);
+ return -ENOMEM;
+ }
+ atomic_inc(&xprt->sc_dma_used);
+
+ n = ib_map_mr_sg(frmr->mr, frmr->sg, frmr->sg_nents, PAGE_SIZE);
+ if (unlikely(n != frmr->sg_nents)) {
+ pr_err("svcrdma: failed to map mr %p (%d/%d elements)\n",
+ frmr->mr, n, frmr->sg_nents);
+ return n < 0 ? n : -EINVAL;
+ }
+
/* Bump the key */
key = (u8)(frmr->mr->lkey & 0x000000FF);
ib_update_fast_reg_key(frmr->mr, ++key);
- ctxt->sge[0].addr = (unsigned long)frmr->kva + *page_offset;
+ ctxt->sge[0].addr = frmr->mr->iova;
ctxt->sge[0].lkey = frmr->mr->lkey;
- ctxt->sge[0].length = read;
+ ctxt->sge[0].length = frmr->mr->length;
ctxt->count = 1;
ctxt->read_hdr = head;
- /* Prepare FASTREG WR */
- memset(&fastreg_wr, 0, sizeof(fastreg_wr));
- fastreg_wr.opcode = IB_WR_FAST_REG_MR;
- fastreg_wr.send_flags = IB_SEND_SIGNALED;
- fastreg_wr.wr.fast_reg.iova_start = (unsigned long)frmr->kva;
- fastreg_wr.wr.fast_reg.page_list = frmr->page_list;
- fastreg_wr.wr.fast_reg.page_list_len = frmr->page_list_len;
- fastreg_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
- fastreg_wr.wr.fast_reg.length = frmr->map_len;
- fastreg_wr.wr.fast_reg.access_flags = frmr->access_flags;
- fastreg_wr.wr.fast_reg.rkey = frmr->mr->lkey;
- fastreg_wr.next = &read_wr;
+ /* Prepare REG WR */
+ reg_wr.wr.opcode = IB_WR_REG_MR;
+ reg_wr.wr.wr_id = 0;
+ reg_wr.wr.send_flags = IB_SEND_SIGNALED;
+ reg_wr.wr.num_sge = 0;
+ reg_wr.mr = frmr->mr;
+ reg_wr.key = frmr->mr->lkey;
+ reg_wr.access = frmr->access_flags;
+ reg_wr.wr.next = &read_wr.wr;
/* Prepare RDMA_READ */
memset(&read_wr, 0, sizeof(read_wr));
- read_wr.send_flags = IB_SEND_SIGNALED;
- read_wr.wr.rdma.rkey = rs_handle;
- read_wr.wr.rdma.remote_addr = rs_offset;
- read_wr.sg_list = ctxt->sge;
- read_wr.num_sge = 1;
+ read_wr.wr.send_flags = IB_SEND_SIGNALED;
+ read_wr.rkey = rs_handle;
+ read_wr.remote_addr = rs_offset;
+ read_wr.wr.sg_list = ctxt->sge;
+ read_wr.wr.num_sge = 1;
if (xprt->sc_dev_caps & SVCRDMA_DEVCAP_READ_W_INV) {
- read_wr.opcode = IB_WR_RDMA_READ_WITH_INV;
- read_wr.wr_id = (unsigned long)ctxt;
- read_wr.ex.invalidate_rkey = ctxt->frmr->mr->lkey;
+ read_wr.wr.opcode = IB_WR_RDMA_READ_WITH_INV;
+ read_wr.wr.wr_id = (unsigned long)ctxt;
+ read_wr.wr.ex.invalidate_rkey = ctxt->frmr->mr->lkey;
} else {
- read_wr.opcode = IB_WR_RDMA_READ;
- read_wr.next = &inv_wr;
+ read_wr.wr.opcode = IB_WR_RDMA_READ;
+ read_wr.wr.next = &inv_wr;
/* Prepare invalidate */
memset(&inv_wr, 0, sizeof(inv_wr));
inv_wr.wr_id = (unsigned long)ctxt;
inv_wr.send_flags = IB_SEND_SIGNALED | IB_SEND_FENCE;
inv_wr.ex.invalidate_rkey = frmr->mr->lkey;
}
- ctxt->wr_op = read_wr.opcode;
+ ctxt->wr_op = read_wr.wr.opcode;
/* Post the chain */
- ret = svc_rdma_send(xprt, &fastreg_wr);
+ ret = svc_rdma_send(xprt, ®_wr.wr);
if (ret) {
pr_err("svcrdma: Error %d posting RDMA_READ\n", ret);
set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags);
atomic_inc(&rdma_stat_read);
return ret;
err:
- svc_rdma_unmap_dma(ctxt);
+ ib_dma_unmap_sg(xprt->sc_cm_id->device,
+ frmr->sg, frmr->sg_nents, frmr->direction);
svc_rdma_put_context(ctxt, 0);
svc_rdma_put_frmr(xprt, frmr);
return ret;
rqstp->rq_arg.page_base = head->arg.page_base;
/* rq_respages starts after the last arg page */
- rqstp->rq_respages = &rqstp->rq_arg.pages[page_no];
+ rqstp->rq_respages = &rqstp->rq_pages[page_no];
rqstp->rq_next_page = rqstp->rq_respages + 1;
/* Rebuild rq_arg head and tail. */
#define RPCDBG_FACILITY RPCDBG_SVCXPRT
+static struct svcxprt_rdma *rdma_create_xprt(struct svc_serv *, int);
static struct svc_xprt *svc_rdma_create(struct svc_serv *serv,
struct net *net,
struct sockaddr *sa, int salen,
.xcl_ident = XPRT_TRANSPORT_RDMA,
};
+#if defined(CONFIG_SUNRPC_BACKCHANNEL)
+static struct svc_xprt *svc_rdma_bc_create(struct svc_serv *, struct net *,
+ struct sockaddr *, int, int);
+static void svc_rdma_bc_detach(struct svc_xprt *);
+static void svc_rdma_bc_free(struct svc_xprt *);
+
+static struct svc_xprt_ops svc_rdma_bc_ops = {
+ .xpo_create = svc_rdma_bc_create,
+ .xpo_detach = svc_rdma_bc_detach,
+ .xpo_free = svc_rdma_bc_free,
+ .xpo_prep_reply_hdr = svc_rdma_prep_reply_hdr,
+ .xpo_secure_port = svc_rdma_secure_port,
+};
+
+struct svc_xprt_class svc_rdma_bc_class = {
+ .xcl_name = "rdma-bc",
+ .xcl_owner = THIS_MODULE,
+ .xcl_ops = &svc_rdma_bc_ops,
+ .xcl_max_payload = (1024 - RPCRDMA_HDRLEN_MIN)
+};
+
+static struct svc_xprt *svc_rdma_bc_create(struct svc_serv *serv,
+ struct net *net,
+ struct sockaddr *sa, int salen,
+ int flags)
+{
+ struct svcxprt_rdma *cma_xprt;
+ struct svc_xprt *xprt;
+
+ cma_xprt = rdma_create_xprt(serv, 0);
+ if (!cma_xprt)
+ return ERR_PTR(-ENOMEM);
+ xprt = &cma_xprt->sc_xprt;
+
+ svc_xprt_init(net, &svc_rdma_bc_class, xprt, serv);
+ serv->sv_bc_xprt = xprt;
+
+ dprintk("svcrdma: %s(%p)\n", __func__, xprt);
+ return xprt;
+}
+
+static void svc_rdma_bc_detach(struct svc_xprt *xprt)
+{
+ dprintk("svcrdma: %s(%p)\n", __func__, xprt);
+}
+
+static void svc_rdma_bc_free(struct svc_xprt *xprt)
+{
+ struct svcxprt_rdma *rdma =
+ container_of(xprt, struct svcxprt_rdma, sc_xprt);
+
+ dprintk("svcrdma: %s(%p)\n", __func__, xprt);
+ if (xprt)
+ kfree(rdma);
+}
+#endif /* CONFIG_SUNRPC_BACKCHANNEL */
+
struct svc_rdma_op_ctxt *svc_rdma_get_context(struct svcxprt_rdma *xprt)
{
struct svc_rdma_op_ctxt *ctxt;
if (!cma_xprt)
return ERR_PTR(-ENOMEM);
- listen_id = rdma_create_id(rdma_listen_handler, cma_xprt, RDMA_PS_TCP,
- IB_QPT_RC);
+ listen_id = rdma_create_id(&init_net, rdma_listen_handler, cma_xprt,
+ RDMA_PS_TCP, IB_QPT_RC);
if (IS_ERR(listen_id)) {
ret = PTR_ERR(listen_id);
dprintk("svcrdma: rdma_create_id failed = %d\n", ret);
static struct svc_rdma_fastreg_mr *rdma_alloc_frmr(struct svcxprt_rdma *xprt)
{
struct ib_mr *mr;
- struct ib_fast_reg_page_list *pl;
+ struct scatterlist *sg;
struct svc_rdma_fastreg_mr *frmr;
u32 num_sg;
if (IS_ERR(mr))
goto err_free_frmr;
- pl = ib_alloc_fast_reg_page_list(xprt->sc_cm_id->device,
- num_sg);
- if (IS_ERR(pl))
+ sg = kcalloc(RPCSVC_MAXPAGES, sizeof(*sg), GFP_KERNEL);
+ if (!sg)
goto err_free_mr;
+ sg_init_table(sg, RPCSVC_MAXPAGES);
+
frmr->mr = mr;
- frmr->page_list = pl;
+ frmr->sg = sg;
INIT_LIST_HEAD(&frmr->frmr_list);
return frmr;
frmr = list_entry(xprt->sc_frmr_q.next,
struct svc_rdma_fastreg_mr, frmr_list);
list_del_init(&frmr->frmr_list);
+ kfree(frmr->sg);
ib_dereg_mr(frmr->mr);
- ib_free_fast_reg_page_list(frmr->page_list);
kfree(frmr);
}
}
frmr = list_entry(rdma->sc_frmr_q.next,
struct svc_rdma_fastreg_mr, frmr_list);
list_del_init(&frmr->frmr_list);
- frmr->map_len = 0;
- frmr->page_list_len = 0;
+ frmr->sg_nents = 0;
}
spin_unlock_bh(&rdma->sc_frmr_q_lock);
if (frmr)
return rdma_alloc_frmr(rdma);
}
- static void frmr_unmap_dma(struct svcxprt_rdma *xprt,
- struct svc_rdma_fastreg_mr *frmr)
- {
- int page_no;
- for (page_no = 0; page_no < frmr->page_list_len; page_no++) {
- dma_addr_t addr = frmr->page_list->page_list[page_no];
- if (ib_dma_mapping_error(frmr->mr->device, addr))
- continue;
- atomic_dec(&xprt->sc_dma_used);
- ib_dma_unmap_page(frmr->mr->device, addr, PAGE_SIZE,
- frmr->direction);
- }
- }
-
void svc_rdma_put_frmr(struct svcxprt_rdma *rdma,
struct svc_rdma_fastreg_mr *frmr)
{
if (frmr) {
- frmr_unmap_dma(rdma, frmr);
+ ib_dma_unmap_sg(rdma->sc_cm_id->device,
+ frmr->sg, frmr->sg_nents, frmr->direction);
+ atomic_dec(&rdma->sc_dma_used);
spin_lock_bh(&rdma->sc_frmr_q_lock);
WARN_ON_ONCE(!list_empty(&frmr->frmr_list));
list_add(&frmr->frmr_list, &rdma->sc_frmr_q);
* internal functions
*/
-/*
- * handle replies in tasklet context, using a single, global list
- * rdma tasklet function -- just turn around and call the func
- * for all replies on the list
- */
-
-static DEFINE_SPINLOCK(rpcrdma_tk_lock_g);
-static LIST_HEAD(rpcrdma_tasklets_g);
+static struct workqueue_struct *rpcrdma_receive_wq;
-static void
-rpcrdma_run_tasklet(unsigned long data)
+int
+rpcrdma_alloc_wq(void)
{
- struct rpcrdma_rep *rep;
- unsigned long flags;
-
- data = data;
- spin_lock_irqsave(&rpcrdma_tk_lock_g, flags);
- while (!list_empty(&rpcrdma_tasklets_g)) {
- rep = list_entry(rpcrdma_tasklets_g.next,
- struct rpcrdma_rep, rr_list);
- list_del(&rep->rr_list);
- spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags);
+ struct workqueue_struct *recv_wq;
- rpcrdma_reply_handler(rep);
+ recv_wq = alloc_workqueue("xprtrdma_receive",
+ WQ_MEM_RECLAIM | WQ_UNBOUND | WQ_HIGHPRI,
+ 0);
+ if (!recv_wq)
+ return -ENOMEM;
- spin_lock_irqsave(&rpcrdma_tk_lock_g, flags);
- }
- spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags);
+ rpcrdma_receive_wq = recv_wq;
+ return 0;
}
-static DECLARE_TASKLET(rpcrdma_tasklet_g, rpcrdma_run_tasklet, 0UL);
-
-static void
-rpcrdma_schedule_tasklet(struct list_head *sched_list)
+void
+rpcrdma_destroy_wq(void)
{
- unsigned long flags;
+ struct workqueue_struct *wq;
- spin_lock_irqsave(&rpcrdma_tk_lock_g, flags);
- list_splice_tail(sched_list, &rpcrdma_tasklets_g);
- spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags);
- tasklet_schedule(&rpcrdma_tasklet_g);
+ if (rpcrdma_receive_wq) {
+ wq = rpcrdma_receive_wq;
+ rpcrdma_receive_wq = NULL;
+ destroy_workqueue(wq);
+ }
}
static void
}
}
-static int
-rpcrdma_sendcq_poll(struct ib_cq *cq, struct rpcrdma_ep *ep)
+/* The common case is a single send completion is waiting. By
+ * passing two WC entries to ib_poll_cq, a return code of 1
+ * means there is exactly one WC waiting and no more. We don't
+ * have to invoke ib_poll_cq again to know that the CQ has been
+ * properly drained.
+ */
+static void
+rpcrdma_sendcq_poll(struct ib_cq *cq)
{
- struct ib_wc *wcs;
- int budget, count, rc;
+ struct ib_wc *pos, wcs[2];
+ int count, rc;
- budget = RPCRDMA_WC_BUDGET / RPCRDMA_POLLSIZE;
do {
- wcs = ep->rep_send_wcs;
+ pos = wcs;
- rc = ib_poll_cq(cq, RPCRDMA_POLLSIZE, wcs);
- if (rc <= 0)
- return rc;
+ rc = ib_poll_cq(cq, ARRAY_SIZE(wcs), pos);
+ if (rc < 0)
+ break;
count = rc;
while (count-- > 0)
- rpcrdma_sendcq_process_wc(wcs++);
- } while (rc == RPCRDMA_POLLSIZE && --budget);
- return 0;
+ rpcrdma_sendcq_process_wc(pos++);
+ } while (rc == ARRAY_SIZE(wcs));
+ return;
}
-/*
- * Handle send, fast_reg_mr, and local_inv completions.
- *
- * Send events are typically suppressed and thus do not result
- * in an upcall. Occasionally one is signaled, however. This
- * prevents the provider's completion queue from wrapping and
- * losing a completion.
+/* Handle provider send completion upcalls.
*/
static void
rpcrdma_sendcq_upcall(struct ib_cq *cq, void *cq_context)
{
- struct rpcrdma_ep *ep = (struct rpcrdma_ep *)cq_context;
- int rc;
-
- rc = rpcrdma_sendcq_poll(cq, ep);
- if (rc) {
- dprintk("RPC: %s: ib_poll_cq failed: %i\n",
- __func__, rc);
- return;
- }
+ do {
+ rpcrdma_sendcq_poll(cq);
+ } while (ib_req_notify_cq(cq, IB_CQ_NEXT_COMP |
+ IB_CQ_REPORT_MISSED_EVENTS) > 0);
+}
- rc = ib_req_notify_cq(cq,
- IB_CQ_NEXT_COMP | IB_CQ_REPORT_MISSED_EVENTS);
- if (rc == 0)
- return;
- if (rc < 0) {
- dprintk("RPC: %s: ib_req_notify_cq failed: %i\n",
- __func__, rc);
- return;
- }
+static void
+rpcrdma_receive_worker(struct work_struct *work)
+{
+ struct rpcrdma_rep *rep =
+ container_of(work, struct rpcrdma_rep, rr_work);
- rpcrdma_sendcq_poll(cq, ep);
+ rpcrdma_reply_handler(rep);
}
static void
-rpcrdma_recvcq_process_wc(struct ib_wc *wc, struct list_head *sched_list)
+rpcrdma_recvcq_process_wc(struct ib_wc *wc)
{
struct rpcrdma_rep *rep =
(struct rpcrdma_rep *)(unsigned long)wc->wr_id;
prefetch(rdmab_to_msg(rep->rr_rdmabuf));
out_schedule:
- list_add_tail(&rep->rr_list, sched_list);
+ queue_work(rpcrdma_receive_wq, &rep->rr_work);
return;
+
out_fail:
if (wc->status != IB_WC_WR_FLUSH_ERR)
pr_err("RPC: %s: rep %p: %s\n",
__func__, rep, ib_wc_status_msg(wc->status));
- rep->rr_len = ~0U;
+ rep->rr_len = RPCRDMA_BAD_LEN;
goto out_schedule;
}
-static int
-rpcrdma_recvcq_poll(struct ib_cq *cq, struct rpcrdma_ep *ep)
+/* The wc array is on stack: automatic memory is always CPU-local.
+ *
+ * struct ib_wc is 64 bytes, making the poll array potentially
+ * large. But this is at the bottom of the call chain. Further
+ * substantial work is done in another thread.
+ */
+static void
+rpcrdma_recvcq_poll(struct ib_cq *cq)
{
- struct list_head sched_list;
- struct ib_wc *wcs;
- int budget, count, rc;
+ struct ib_wc *pos, wcs[4];
+ int count, rc;
- INIT_LIST_HEAD(&sched_list);
- budget = RPCRDMA_WC_BUDGET / RPCRDMA_POLLSIZE;
do {
- wcs = ep->rep_recv_wcs;
+ pos = wcs;
- rc = ib_poll_cq(cq, RPCRDMA_POLLSIZE, wcs);
- if (rc <= 0)
- goto out_schedule;
+ rc = ib_poll_cq(cq, ARRAY_SIZE(wcs), pos);
+ if (rc < 0)
+ break;
count = rc;
while (count-- > 0)
- rpcrdma_recvcq_process_wc(wcs++, &sched_list);
- } while (rc == RPCRDMA_POLLSIZE && --budget);
- rc = 0;
-
-out_schedule:
- rpcrdma_schedule_tasklet(&sched_list);
- return rc;
+ rpcrdma_recvcq_process_wc(pos++);
+ } while (rc == ARRAY_SIZE(wcs));
}
-/*
- * Handle receive completions.
- *
- * It is reentrant but processes single events in order to maintain
- * ordering of receives to keep server credits.
- *
- * It is the responsibility of the scheduled tasklet to return
- * recv buffers to the pool. NOTE: this affects synchronization of
- * connection shutdown. That is, the structures required for
- * the completion of the reply handler must remain intact until
- * all memory has been reclaimed.
+/* Handle provider receive completion upcalls.
*/
static void
rpcrdma_recvcq_upcall(struct ib_cq *cq, void *cq_context)
{
- struct rpcrdma_ep *ep = (struct rpcrdma_ep *)cq_context;
- int rc;
-
- rc = rpcrdma_recvcq_poll(cq, ep);
- if (rc) {
- dprintk("RPC: %s: ib_poll_cq failed: %i\n",
- __func__, rc);
- return;
- }
-
- rc = ib_req_notify_cq(cq,
- IB_CQ_NEXT_COMP | IB_CQ_REPORT_MISSED_EVENTS);
- if (rc == 0)
- return;
- if (rc < 0) {
- dprintk("RPC: %s: ib_req_notify_cq failed: %i\n",
- __func__, rc);
- return;
- }
-
- rpcrdma_recvcq_poll(cq, ep);
+ do {
+ rpcrdma_recvcq_poll(cq);
+ } while (ib_req_notify_cq(cq, IB_CQ_NEXT_COMP |
+ IB_CQ_REPORT_MISSED_EVENTS) > 0);
}
static void
rpcrdma_flush_cqs(struct rpcrdma_ep *ep)
{
struct ib_wc wc;
- LIST_HEAD(sched_list);
while (ib_poll_cq(ep->rep_attr.recv_cq, 1, &wc) > 0)
- rpcrdma_recvcq_process_wc(&wc, &sched_list);
- if (!list_empty(&sched_list))
- rpcrdma_schedule_tasklet(&sched_list);
+ rpcrdma_recvcq_process_wc(&wc);
while (ib_poll_cq(ep->rep_attr.send_cq, 1, &wc) > 0)
rpcrdma_sendcq_process_wc(&wc);
}
init_completion(&ia->ri_done);
- id = rdma_create_id(rpcrdma_conn_upcall, xprt, RDMA_PS_TCP, IB_QPT_RC);
+ id = rdma_create_id(&init_net, rpcrdma_conn_upcall, xprt, RDMA_PS_TCP,
+ IB_QPT_RC);
if (IS_ERR(id)) {
rc = PTR_ERR(id);
dprintk("RPC: %s: rdma_create_id() failed %i\n",
struct ib_device_attr *devattr = &ia->ri_devattr;
struct ib_cq *sendcq, *recvcq;
struct ib_cq_init_attr cq_attr = {};
+ unsigned int max_qp_wr;
int rc, err;
if (devattr->max_sge < RPCRDMA_MAX_IOVS) {
return -ENOMEM;
}
+ if (devattr->max_qp_wr <= RPCRDMA_BACKWARD_WRS) {
+ dprintk("RPC: %s: insufficient wqe's available\n",
+ __func__);
+ return -ENOMEM;
+ }
+ max_qp_wr = devattr->max_qp_wr - RPCRDMA_BACKWARD_WRS;
+
/* check provider's send/recv wr limits */
- if (cdata->max_requests > devattr->max_qp_wr)
- cdata->max_requests = devattr->max_qp_wr;
+ if (cdata->max_requests > max_qp_wr)
+ cdata->max_requests = max_qp_wr;
ep->rep_attr.event_handler = rpcrdma_qp_async_error_upcall;
ep->rep_attr.qp_context = ep;
ep->rep_attr.srq = NULL;
ep->rep_attr.cap.max_send_wr = cdata->max_requests;
+ ep->rep_attr.cap.max_send_wr += RPCRDMA_BACKWARD_WRS;
rc = ia->ri_ops->ro_open(ia, ep, cdata);
if (rc)
return rc;
ep->rep_attr.cap.max_recv_wr = cdata->max_requests;
+ ep->rep_attr.cap.max_recv_wr += RPCRDMA_BACKWARD_WRS;
ep->rep_attr.cap.max_send_sge = RPCRDMA_MAX_IOVS;
ep->rep_attr.cap.max_recv_sge = 1;
ep->rep_attr.cap.max_inline_data = 0;
cq_attr.cqe = ep->rep_attr.cap.max_send_wr + 1;
sendcq = ib_create_cq(ia->ri_device, rpcrdma_sendcq_upcall,
- rpcrdma_cq_async_error_upcall, ep, &cq_attr);
+ rpcrdma_cq_async_error_upcall, NULL, &cq_attr);
if (IS_ERR(sendcq)) {
rc = PTR_ERR(sendcq);
dprintk("RPC: %s: failed to create send CQ: %i\n",
cq_attr.cqe = ep->rep_attr.cap.max_recv_wr + 1;
recvcq = ib_create_cq(ia->ri_device, rpcrdma_recvcq_upcall,
- rpcrdma_cq_async_error_upcall, ep, &cq_attr);
+ rpcrdma_cq_async_error_upcall, NULL, &cq_attr);
if (IS_ERR(recvcq)) {
rc = PTR_ERR(recvcq);
dprintk("RPC: %s: failed to create recv CQ: %i\n",
cancel_delayed_work_sync(&ep->rep_connect_worker);
- if (ia->ri_id->qp) {
+ if (ia->ri_id->qp)
rpcrdma_ep_disconnect(ep, ia);
+
+ rpcrdma_clean_cq(ep->rep_attr.recv_cq);
+ rpcrdma_clean_cq(ep->rep_attr.send_cq);
+
+ if (ia->ri_id->qp) {
rdma_destroy_qp(ia->ri_id);
ia->ri_id->qp = NULL;
}
- rpcrdma_clean_cq(ep->rep_attr.recv_cq);
rc = ib_destroy_cq(ep->rep_attr.recv_cq);
if (rc)
dprintk("RPC: %s: ib_destroy_cq returned %i\n",
__func__, rc);
- rpcrdma_clean_cq(ep->rep_attr.send_cq);
rc = ib_destroy_cq(ep->rep_attr.send_cq);
if (rc)
dprintk("RPC: %s: ib_destroy_cq returned %i\n",
}
rc = ep->rep_connected;
} else {
+ struct rpcrdma_xprt *r_xprt;
+ unsigned int extras;
+
dprintk("RPC: %s: connected\n", __func__);
+
+ r_xprt = container_of(ia, struct rpcrdma_xprt, rx_ia);
+ extras = r_xprt->rx_buf.rb_bc_srv_max_requests;
+
+ if (extras) {
+ rc = rpcrdma_ep_post_extra_recv(r_xprt, extras);
+ if (rc)
+ pr_warn("%s: rpcrdma_ep_post_extra_recv: %i\n",
+ __func__, rc);
+ rc = 0;
+ }
}
out:
}
}
-static struct rpcrdma_req *
+struct rpcrdma_req *
rpcrdma_create_req(struct rpcrdma_xprt *r_xprt)
{
+ struct rpcrdma_buffer *buffer = &r_xprt->rx_buf;
struct rpcrdma_req *req;
req = kzalloc(sizeof(*req), GFP_KERNEL);
if (req == NULL)
return ERR_PTR(-ENOMEM);
+ INIT_LIST_HEAD(&req->rl_free);
+ spin_lock(&buffer->rb_reqslock);
+ list_add(&req->rl_all, &buffer->rb_allreqs);
+ spin_unlock(&buffer->rb_reqslock);
req->rl_buffer = &r_xprt->rx_buf;
return req;
}
-static struct rpcrdma_rep *
+struct rpcrdma_rep *
rpcrdma_create_rep(struct rpcrdma_xprt *r_xprt)
{
struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
rep->rr_device = ia->ri_device;
rep->rr_rxprt = r_xprt;
+ INIT_WORK(&rep->rr_work, rpcrdma_receive_worker);
return rep;
out_free:
{
struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
struct rpcrdma_ia *ia = &r_xprt->rx_ia;
- struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
- char *p;
- size_t len;
int i, rc;
- buf->rb_max_requests = cdata->max_requests;
+ buf->rb_max_requests = r_xprt->rx_data.max_requests;
+ buf->rb_bc_srv_max_requests = 0;
spin_lock_init(&buf->rb_lock);
- /* Need to allocate:
- * 1. arrays for send and recv pointers
- * 2. arrays of struct rpcrdma_req to fill in pointers
- * 3. array of struct rpcrdma_rep for replies
- * Send/recv buffers in req/rep need to be registered
- */
- len = buf->rb_max_requests *
- (sizeof(struct rpcrdma_req *) + sizeof(struct rpcrdma_rep *));
-
- p = kzalloc(len, GFP_KERNEL);
- if (p == NULL) {
- dprintk("RPC: %s: req_t/rep_t/pad kzalloc(%zd) failed\n",
- __func__, len);
- rc = -ENOMEM;
- goto out;
- }
- buf->rb_pool = p; /* for freeing it later */
-
- buf->rb_send_bufs = (struct rpcrdma_req **) p;
- p = (char *) &buf->rb_send_bufs[buf->rb_max_requests];
- buf->rb_recv_bufs = (struct rpcrdma_rep **) p;
- p = (char *) &buf->rb_recv_bufs[buf->rb_max_requests];
-
rc = ia->ri_ops->ro_init(r_xprt);
if (rc)
goto out;
+ INIT_LIST_HEAD(&buf->rb_send_bufs);
+ INIT_LIST_HEAD(&buf->rb_allreqs);
+ spin_lock_init(&buf->rb_reqslock);
for (i = 0; i < buf->rb_max_requests; i++) {
struct rpcrdma_req *req;
- struct rpcrdma_rep *rep;
req = rpcrdma_create_req(r_xprt);
if (IS_ERR(req)) {
rc = PTR_ERR(req);
goto out;
}
- buf->rb_send_bufs[i] = req;
+ req->rl_backchannel = false;
+ list_add(&req->rl_free, &buf->rb_send_bufs);
+ }
+
+ INIT_LIST_HEAD(&buf->rb_recv_bufs);
+ for (i = 0; i < buf->rb_max_requests + 2; i++) {
+ struct rpcrdma_rep *rep;
rep = rpcrdma_create_rep(r_xprt);
if (IS_ERR(rep)) {
rc = PTR_ERR(rep);
goto out;
}
- buf->rb_recv_bufs[i] = rep;
+ list_add(&rep->rr_list, &buf->rb_recv_bufs);
}
return 0;
return rc;
}
+static struct rpcrdma_req *
+rpcrdma_buffer_get_req_locked(struct rpcrdma_buffer *buf)
+{
+ struct rpcrdma_req *req;
+
+ req = list_first_entry(&buf->rb_send_bufs,
+ struct rpcrdma_req, rl_free);
+ list_del(&req->rl_free);
+ return req;
+}
+
+static struct rpcrdma_rep *
+rpcrdma_buffer_get_rep_locked(struct rpcrdma_buffer *buf)
+{
+ struct rpcrdma_rep *rep;
+
+ rep = list_first_entry(&buf->rb_recv_bufs,
+ struct rpcrdma_rep, rr_list);
+ list_del(&rep->rr_list);
+ return rep;
+}
+
static void
rpcrdma_destroy_rep(struct rpcrdma_ia *ia, struct rpcrdma_rep *rep)
{
- if (!rep)
- return;
-
rpcrdma_free_regbuf(ia, rep->rr_rdmabuf);
kfree(rep);
}
-static void
+void
rpcrdma_destroy_req(struct rpcrdma_ia *ia, struct rpcrdma_req *req)
{
- if (!req)
- return;
-
rpcrdma_free_regbuf(ia, req->rl_sendbuf);
rpcrdma_free_regbuf(ia, req->rl_rdmabuf);
kfree(req);
rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf)
{
struct rpcrdma_ia *ia = rdmab_to_ia(buf);
- int i;
- /* clean up in reverse order from create
- * 1. recv mr memory (mr free, then kfree)
- * 2. send mr memory (mr free, then kfree)
- * 3. MWs
- */
- dprintk("RPC: %s: entering\n", __func__);
+ while (!list_empty(&buf->rb_recv_bufs)) {
+ struct rpcrdma_rep *rep;
- for (i = 0; i < buf->rb_max_requests; i++) {
- if (buf->rb_recv_bufs)
- rpcrdma_destroy_rep(ia, buf->rb_recv_bufs[i]);
- if (buf->rb_send_bufs)
- rpcrdma_destroy_req(ia, buf->rb_send_bufs[i]);
+ rep = rpcrdma_buffer_get_rep_locked(buf);
+ rpcrdma_destroy_rep(ia, rep);
}
- ia->ri_ops->ro_destroy(buf);
+ spin_lock(&buf->rb_reqslock);
+ while (!list_empty(&buf->rb_allreqs)) {
+ struct rpcrdma_req *req;
- kfree(buf->rb_pool);
+ req = list_first_entry(&buf->rb_allreqs,
+ struct rpcrdma_req, rl_all);
+ list_del(&req->rl_all);
+
+ spin_unlock(&buf->rb_reqslock);
+ rpcrdma_destroy_req(ia, req);
+ spin_lock(&buf->rb_reqslock);
+ }
+ spin_unlock(&buf->rb_reqslock);
+
+ ia->ri_ops->ro_destroy(buf);
}
struct rpcrdma_mw *
spin_unlock(&buf->rb_mwlock);
}
-static void
-rpcrdma_buffer_put_sendbuf(struct rpcrdma_req *req, struct rpcrdma_buffer *buf)
-{
- buf->rb_send_bufs[--buf->rb_send_index] = req;
- req->rl_niovs = 0;
- if (req->rl_reply) {
- buf->rb_recv_bufs[--buf->rb_recv_index] = req->rl_reply;
- req->rl_reply = NULL;
- }
-}
-
/*
* Get a set of request/reply buffers.
*
- * Reply buffer (if needed) is attached to send buffer upon return.
- * Rule:
- * rb_send_index and rb_recv_index MUST always be pointing to the
- * *next* available buffer (non-NULL). They are incremented after
- * removing buffers, and decremented *before* returning them.
+ * Reply buffer (if available) is attached to send buffer upon return.
*/
struct rpcrdma_req *
rpcrdma_buffer_get(struct rpcrdma_buffer *buffers)
{
struct rpcrdma_req *req;
- unsigned long flags;
-
- spin_lock_irqsave(&buffers->rb_lock, flags);
-
- if (buffers->rb_send_index == buffers->rb_max_requests) {
- spin_unlock_irqrestore(&buffers->rb_lock, flags);
- dprintk("RPC: %s: out of request buffers\n", __func__);
- return ((struct rpcrdma_req *)NULL);
- }
- req = buffers->rb_send_bufs[buffers->rb_send_index];
- if (buffers->rb_send_index < buffers->rb_recv_index) {
- dprintk("RPC: %s: %d extra receives outstanding (ok)\n",
- __func__,
- buffers->rb_recv_index - buffers->rb_send_index);
- req->rl_reply = NULL;
- } else {
- req->rl_reply = buffers->rb_recv_bufs[buffers->rb_recv_index];
- buffers->rb_recv_bufs[buffers->rb_recv_index++] = NULL;
- }
- buffers->rb_send_bufs[buffers->rb_send_index++] = NULL;
+ spin_lock(&buffers->rb_lock);
+ if (list_empty(&buffers->rb_send_bufs))
+ goto out_reqbuf;
+ req = rpcrdma_buffer_get_req_locked(buffers);
+ if (list_empty(&buffers->rb_recv_bufs))
+ goto out_repbuf;
+ req->rl_reply = rpcrdma_buffer_get_rep_locked(buffers);
+ spin_unlock(&buffers->rb_lock);
+ return req;
- spin_unlock_irqrestore(&buffers->rb_lock, flags);
+out_reqbuf:
+ spin_unlock(&buffers->rb_lock);
+ pr_warn("RPC: %s: out of request buffers\n", __func__);
+ return NULL;
+out_repbuf:
+ spin_unlock(&buffers->rb_lock);
+ pr_warn("RPC: %s: out of reply buffers\n", __func__);
+ req->rl_reply = NULL;
return req;
}
rpcrdma_buffer_put(struct rpcrdma_req *req)
{
struct rpcrdma_buffer *buffers = req->rl_buffer;
- unsigned long flags;
+ struct rpcrdma_rep *rep = req->rl_reply;
- spin_lock_irqsave(&buffers->rb_lock, flags);
- rpcrdma_buffer_put_sendbuf(req, buffers);
- spin_unlock_irqrestore(&buffers->rb_lock, flags);
+ req->rl_niovs = 0;
+ req->rl_reply = NULL;
+
+ spin_lock(&buffers->rb_lock);
+ list_add_tail(&req->rl_free, &buffers->rb_send_bufs);
+ if (rep)
+ list_add_tail(&rep->rr_list, &buffers->rb_recv_bufs);
+ spin_unlock(&buffers->rb_lock);
}
/*
* Recover reply buffers from pool.
- * This happens when recovering from error conditions.
- * Post-increment counter/array index.
+ * This happens when recovering from disconnect.
*/
void
rpcrdma_recv_buffer_get(struct rpcrdma_req *req)
{
struct rpcrdma_buffer *buffers = req->rl_buffer;
- unsigned long flags;
- spin_lock_irqsave(&buffers->rb_lock, flags);
- if (buffers->rb_recv_index < buffers->rb_max_requests) {
- req->rl_reply = buffers->rb_recv_bufs[buffers->rb_recv_index];
- buffers->rb_recv_bufs[buffers->rb_recv_index++] = NULL;
- }
- spin_unlock_irqrestore(&buffers->rb_lock, flags);
+ spin_lock(&buffers->rb_lock);
+ if (!list_empty(&buffers->rb_recv_bufs))
+ req->rl_reply = rpcrdma_buffer_get_rep_locked(buffers);
+ spin_unlock(&buffers->rb_lock);
}
/*
rpcrdma_recv_buffer_put(struct rpcrdma_rep *rep)
{
struct rpcrdma_buffer *buffers = &rep->rr_rxprt->rx_buf;
- unsigned long flags;
- spin_lock_irqsave(&buffers->rb_lock, flags);
- buffers->rb_recv_bufs[--buffers->rb_recv_index] = rep;
- spin_unlock_irqrestore(&buffers->rb_lock, flags);
+ spin_lock(&buffers->rb_lock);
+ list_add_tail(&rep->rr_list, &buffers->rb_recv_bufs);
+ spin_unlock(&buffers->rb_lock);
}
/*
return rc;
}
+/**
+ * rpcrdma_ep_post_extra_recv - Post buffers for incoming backchannel requests
+ * @r_xprt: transport associated with these backchannel resources
+ * @min_reqs: minimum number of incoming requests expected
+ *
+ * Returns zero if all requested buffers were posted, or a negative errno.
+ */
+int
+rpcrdma_ep_post_extra_recv(struct rpcrdma_xprt *r_xprt, unsigned int count)
+{
+ struct rpcrdma_buffer *buffers = &r_xprt->rx_buf;
+ struct rpcrdma_ia *ia = &r_xprt->rx_ia;
+ struct rpcrdma_ep *ep = &r_xprt->rx_ep;
+ struct rpcrdma_rep *rep;
+ unsigned long flags;
+ int rc;
+
+ while (count--) {
+ spin_lock_irqsave(&buffers->rb_lock, flags);
+ if (list_empty(&buffers->rb_recv_bufs))
+ goto out_reqbuf;
+ rep = rpcrdma_buffer_get_rep_locked(buffers);
+ spin_unlock_irqrestore(&buffers->rb_lock, flags);
+
+ rc = rpcrdma_ep_post_recv(ia, ep, rep);
+ if (rc)
+ goto out_rc;
+ }
+
+ return 0;
+
+out_reqbuf:
+ spin_unlock_irqrestore(&buffers->rb_lock, flags);
+ pr_warn("%s: no extra receive buffers\n", __func__);
+ return -ENOMEM;
+
+out_rc:
+ rpcrdma_recv_buffer_put(rep);
+ return rc;
+}
+
/* How many chunk list items fit within our inline buffers?
*/
unsigned int
* RDMA Endpoint -- one per transport instance
*/
-#define RPCRDMA_WC_BUDGET (128)
-#define RPCRDMA_POLLSIZE (16)
-
struct rpcrdma_ep {
atomic_t rep_cqcount;
int rep_cqinit;
struct rdma_conn_param rep_remote_cma;
struct sockaddr_storage rep_remote_addr;
struct delayed_work rep_connect_worker;
- struct ib_wc rep_send_wcs[RPCRDMA_POLLSIZE];
- struct ib_wc rep_recv_wcs[RPCRDMA_POLLSIZE];
};
/*
*/
#define RPCRDMA_IGNORE_COMPLETION (0ULL)
+/* Pre-allocate extra Work Requests for handling backward receives
+ * and sends. This is a fixed value because the Work Queues are
+ * allocated when the forward channel is set up.
+ */
+#if defined(CONFIG_SUNRPC_BACKCHANNEL)
+#define RPCRDMA_BACKWARD_WRS (8)
+#else
+#define RPCRDMA_BACKWARD_WRS (0)
+#endif
+
/* Registered buffer -- registered kmalloc'd memory for RDMA SEND/RECV
*
* The below structure appears at the front of a large region of kmalloc'd
unsigned int rr_len;
struct ib_device *rr_device;
struct rpcrdma_xprt *rr_rxprt;
+ struct work_struct rr_work;
struct list_head rr_list;
struct rpcrdma_regbuf *rr_rdmabuf;
};
+#define RPCRDMA_BAD_LEN (~0U)
+
/*
* struct rpcrdma_mw - external memory region metadata
*
};
struct rpcrdma_frmr {
- struct ib_fast_reg_page_list *fr_pgl;
+ struct scatterlist *sg;
+ int sg_nents;
struct ib_mr *fr_mr;
enum rpcrdma_frmr_state fr_state;
struct work_struct fr_work;
#define RPCRDMA_MAX_IOVS (2)
struct rpcrdma_req {
+ struct list_head rl_free;
unsigned int rl_niovs;
unsigned int rl_nchunks;
unsigned int rl_connect_cookie;
struct rpcrdma_regbuf *rl_rdmabuf;
struct rpcrdma_regbuf *rl_sendbuf;
struct rpcrdma_mr_seg rl_segments[RPCRDMA_MAX_SEGS];
+
+ struct list_head rl_all;
+ bool rl_backchannel;
};
static inline struct rpcrdma_req *
struct list_head rb_all;
char *rb_pool;
- spinlock_t rb_lock; /* protect buf arrays */
+ spinlock_t rb_lock; /* protect buf lists */
+ struct list_head rb_send_bufs;
+ struct list_head rb_recv_bufs;
u32 rb_max_requests;
- int rb_send_index;
- int rb_recv_index;
- struct rpcrdma_req **rb_send_bufs;
- struct rpcrdma_rep **rb_recv_bufs;
+
+ u32 rb_bc_srv_max_requests;
+ spinlock_t rb_reqslock; /* protect rb_allreqs */
+ struct list_head rb_allreqs;
};
#define rdmab_to_ia(b) (&container_of((b), struct rpcrdma_xprt, rx_buf)->rx_ia)
unsigned long failed_marshal_count;
unsigned long bad_reply_count;
unsigned long nomsg_call_count;
+ unsigned long bcall_count;
};
/*
/*
* Buffer calls - xprtrdma/verbs.c
*/
+struct rpcrdma_req *rpcrdma_create_req(struct rpcrdma_xprt *);
+struct rpcrdma_rep *rpcrdma_create_rep(struct rpcrdma_xprt *);
+void rpcrdma_destroy_req(struct rpcrdma_ia *, struct rpcrdma_req *);
int rpcrdma_buffer_create(struct rpcrdma_xprt *);
void rpcrdma_buffer_destroy(struct rpcrdma_buffer *);
struct rpcrdma_regbuf *);
unsigned int rpcrdma_max_segments(struct rpcrdma_xprt *);
+int rpcrdma_ep_post_extra_recv(struct rpcrdma_xprt *, unsigned int);
int frwr_alloc_recovery_wq(void);
void frwr_destroy_recovery_wq(void);
+int rpcrdma_alloc_wq(void);
+void rpcrdma_destroy_wq(void);
+
/*
* Wrappers for chunk registration, shared by read/write chunk code.
*/
int xprt_rdma_init(void);
void xprt_rdma_cleanup(void);
+/* Backchannel calls - xprtrdma/backchannel.c
+ */
+#if defined(CONFIG_SUNRPC_BACKCHANNEL)
+int xprt_rdma_bc_setup(struct rpc_xprt *, unsigned int);
+int xprt_rdma_bc_up(struct svc_serv *, struct net *);
+int rpcrdma_bc_post_recv(struct rpcrdma_xprt *, unsigned int);
+void rpcrdma_bc_receive_call(struct rpcrdma_xprt *, struct rpcrdma_rep *);
+int rpcrdma_bc_marshal_reply(struct rpc_rqst *);
+void xprt_rdma_bc_free_rqst(struct rpc_rqst *);
+void xprt_rdma_bc_destroy(struct rpc_xprt *, unsigned int);
+#endif /* CONFIG_SUNRPC_BACKCHANNEL */
+
/* Temporary NFS request map cache. Created in svc_rdma.c */
extern struct kmem_cache *svc_rdma_map_cachep;
/* WR context cache. Created in svc_rdma.c */
projects / karo-tx-linux.git / commitdiff