SeaBIOS October 2013

seabios@seabios.org

14 participants
51 discussions

QEMU detect and PCI Configuration Space
by SeaBIOS 15 Oct '13

15 Oct '13

Dear all, I am reading the source code of SeaBIOS. In QEMU detection part (qemu_detect()), I notice the method to detect QEMU is the same as detect other PCI devices through PCI Configuration Space (I/O port: 0xcf8 and 0xcfc). Does QEMU emulate the PCI Configuration Space? Any help or comments will be appreciated. Thank you Best, Min Li

2 1

[PATCH] Build different final files for QEMU, coreboot, and CSM.
by Kevin O'Connor 15 Oct '13

15 Oct '13

Build out/bios.bin on QEMU, out/bios.bin.elf on coreboot, and out/Csm16.bin for CSM. This reduces the chance that one accidentally builds for an incorrect target. Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- Makefile | 20 +++++++++++++++++--- README | 6 ++++-- README.CSM | 11 ++++++----- 3 files changed, 27 insertions(+), 10 deletions(-) diff --git a/Makefile b/Makefile index 01bd8b3..b1b55a3 100644 --- a/Makefile +++ b/Makefile @@ -82,7 +82,10 @@ endif # Default targets -include $(KCONFIG_CONFIG) -target-y = $(OUT)bios.bin +target-y := +target-$(CONFIG_QEMU) += $(OUT)bios.bin +target-$(CONFIG_CSM) += $(OUT)Csm16.bin +target-$(CONFIG_COREBOOT) += $(OUT)bios.bin.elf target-$(CONFIG_BUILD_VGABIOS) += $(OUT)vgabios.bin all: $(target-y) @@ -174,12 +177,23 @@ $(OUT)rom.o: $(OUT)rom16.strip.o $(OUT)rom32seg.strip.o $(OUT)code32flat.o $(OUT @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout32flat.lds $(OUT)rom16.strip.o $(OUT)rom32seg.strip.o $(OUT)code32flat.o -o $@ -$(OUT)bios.bin.elf $(OUT)bios.bin: $(OUT)rom.o scripts/checkrom.py +$(OUT)bios.bin.prep: $(OUT)rom.o scripts/checkrom.py @echo " Prepping $@" $(Q)$(OBJDUMP) -thr $< > $<.objdump $(Q)$(OBJCOPY) -O binary $< $(OUT)bios.bin.raw $(Q)$(PYTHON) ./scripts/checkrom.py $<.objdump $(CONFIG_ROM_SIZE) \ - $(OUT)bios.bin.raw $(OUT)bios.bin + $(OUT)bios.bin.raw $(OUT)bios.bin.prep + +$(OUT)bios.bin: $(OUT)bios.bin.prep + @echo " Creating $@" + $(Q)cp $< $@ + +$(OUT)Csm16.bin: $(OUT)bios.bin.prep + @echo " Creating $@" + $(Q)cp $< $@ + +$(OUT)bios.bin.elf: $(OUT)rom.o $(OUT)bios.bin.prep + @echo " Creating $@" $(Q)$(STRIP) -R .comment $< -o $(OUT)bios.bin.elf diff --git a/README b/README index f752451..6844e52 100644 --- a/README +++ b/README @@ -1,8 +1,10 @@ This code implements an X86 legacy bios. It is intended to be compiled using standard gnu tools (eg, gas and gcc). -To build, one should be able to run "make" in the main directory. The -resulting file "out/bios.bin" contains the processed bios image. +To build for QEMU, one should be able to run "make" in the main +directory. The resulting file "out/bios.bin" contains the processed +bios image. To build for coreboot, please see the coreboot wiki. To +build for CSM, please see README.CSM. Testing of images: diff --git a/README.CSM b/README.CSM index ea71846..b904e65 100644 --- a/README.CSM +++ b/README.CSM @@ -9,11 +9,12 @@ Windows 2008r2 is known to use INT 10h BIOS calls even when booted via EFI, and the presence of a CSM makes this work as expected too. Having built SeaBIOS with CONFIG_CSM, you should be able to drop the -result into your OVMF build tree at OvmfPkg/Csm/Csm16/Csm16.bin and -then build OVMF with 'build -D CSM_ENABLE'. The SeaBIOS binary will be -included as a discrete file within the 'Flash Volume' which is -created, and there are tools which will extract it and allow it to be -replaced; satisfying the requirements of the LGPL licence. +result (out/Csm16.bin) into your OVMF build tree at +OvmfPkg/Csm/Csm16/Csm16.bin and then build OVMF with 'build -D +CSM_ENABLE'. The SeaBIOS binary will be included as a discrete file +within the 'Flash Volume' which is created, and there are tools which +will extract it and allow it to be replaced; satisfying the +requirements of the LGPL licence. A patch to OVMF is required, to prevent it from marking the region from 0xC0000-0xFFFFF as read-only before invoking our Legacy16Boot method. See -- 1.8.3.1

1 0

[PATCH 0/6] Convert AHCI driver to run entirely in 32bit mode
by Kevin O'Connor 15 Oct '13

15 Oct '13

This series converts the AHCI driver to run entirely in 32bit mode. Currently, ahci will only jump into 32bit mode when it needs to access the ahci device's PCI config space. However, if the code has to run anything in 32bit mode, it might as well run it all in 32bit mode. Doing this also shrinks the total SeaBIOS code size by a few KB. In order for this to work, SeaBIOS needs to be able to transition from 32bit mode back to 16bit mode in order to use yield(). This requires tracking the stack segment when using call32() and restoring it if call16() is later called. It also requires getting off the "extra stack" if it is in use. This series is a little bit uglier than I hoped, but I'm sending it out for further comments. This series is also at: https://github.com/KevinOConnor/seabios/tree/testing -Kevin Kevin O'Connor (6): Rearrange stack_hop_back() call in wait_irq, check_irqs, and _farcall16. Minor - move call16 assembler in romlayout.S. Make __call16 use C calling convention and support two passed parameters. Update _farcall16() to pass segment of callregs explicitly. Support call16() calls after entering 32bit mode from call32(). Run ahci code entirely in 32bit mode. Makefile | 4 +- src/block.c | 13 ++++-- src/hw/ahci.c | 15 ++----- src/hw/blockcmd.c | 5 ++- src/romlayout.S | 132 ++++++++++++++++++++++++++++++++---------------------- src/stacks.c | 78 +++++++++++++++++++------------- 6 files changed, 145 insertions(+), 102 deletions(-) -- 1.8.3.1

2 3

Re: [SeaBIOS] [PATCH v4] Add pvscsi boot support
by Paul Menzel 15 Oct '13

15 Oct '13

Dear Evgeny, thanks again for the next iteration! Am Montag, den 14.10.2013, 18:03 +0300 schrieb Evgeny Budilovsky: > Testing was done on windows images (win 2008/2012) taken from esx with vmware > tools installed and boot disk configured to use pvscsi. > Also I've used linux (ubuntu 12.04) where pvscsi drivers are installed by > default and booted it using qemu cmd similar to this: > > ./x86_64-softmmu/qemu-system-x86_64 -snapshot -device pvscsi,id=pvscsi0 \ > -device scsi-disk,bus=pvscsi0.0,drive=drive0 \ > -drive id=drive0,if=none,file=ubuntu-12.04.qcow2 \ > -bios roms/seabios/out/bios.bin > > Signed-off-by: Evgeny Budilovsky <evgeny.budilovsky(a)ravellosystems.com> > --- > Makefile | 2 +- > src/Kconfig | 12 ++ > src/block.c | 1 + > src/block.h | 1 + > src/hw/blockcmd.c | 3 + > src/hw/pci_ids.h | 3 + > src/hw/pvscsi.c | 363 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > src/hw/pvscsi.h | 8 ++ > src/post.c | 2 + > 9 files changed, 394 insertions(+), 1 deletion(-) > create mode 100644 src/hw/pvscsi.c > create mode 100644 src/hw/pvscsi.h […] > diff --git a/src/hw/pvscsi.c b/src/hw/pvscsi.c > new file mode 100644 > index 0000000..740a74b > --- /dev/null > +++ b/src/hw/pvscsi.c > @@ -0,0 +1,363 @@ > +// QEMU VMWARE Paravirtualized SCSI boot support. > +// > +// Copyright (c) 2013 Ravello Systems LTD (http://ravellosystems.com) > +// > +// Authors: > +// Evgeny Budilovsky <evgeny.budilovsky(a)ravellosystems.com> > +// > +// This file may be distributed under the terms of the GNU LGPLv3 license. > + > +#include "biosvar.h" // GET_GLOBAL > +#include "block.h" // struct drive_s > +#include "blockcmd.h" // scsi_drive_setup > +#include "config.h" // CONFIG_* > +#include "malloc.h" // free > +#include "output.h" // dprintf > +#include "pci.h" // foreachpci > +#include "pci_ids.h" // PCI_DEVICE_ID_VMWARE_PVSCSI > +#include "pci_regs.h" // PCI_VENDOR_ID > +#include "std/disk.h" // DISK_RET_SUCCESS > +#include "string.h" // memset > +#include "util.h" // usleep > +#include "pvscsi.h" > +#include "virtio-ring.h" // PAGE_SHIFT, virt_to_phys > + > +#define MASK(n) ((1 << (n)) - 1) > + > +#define SIMPLE_QUEUE_TAG 0x20 > + > +#define PVSCSI_INTR_CMPL_0 (1 << 0) > +#define PVSCSI_INTR_CMPL_1 (1 << 1) > +#define PVSCSI_INTR_CMPL_MASK MASK(2) > + > +#define PVSCSI_INTR_MSG_0 (1 << 2) > +#define PVSCSI_INTR_MSG_1 (1 << 3) > +#define PVSCSI_INTR_MSG_MASK (MASK(2) << 2) > +#define PVSCSI_INTR_ALL_SUPPORTED MASK(4) > + > +#define PVSCSI_FLAG_CMD_WITH_SG_LIST (1 << 0) > +#define PVSCSI_FLAG_CMD_OUT_OF_BAND_CDB (1 << 1) > +#define PVSCSI_FLAG_CMD_DIR_NONE (1 << 2) > +#define PVSCSI_FLAG_CMD_DIR_TOHOST (1 << 3) > +#define PVSCSI_FLAG_CMD_DIR_TODEVICE (1 << 4) > + > +enum PVSCSIRegOffset { > + PVSCSI_REG_OFFSET_COMMAND = 0x0, > + PVSCSI_REG_OFFSET_COMMAND_DATA = 0x4, > + PVSCSI_REG_OFFSET_COMMAND_STATUS = 0x8, > + PVSCSI_REG_OFFSET_LAST_STS_0 = 0x100, > + PVSCSI_REG_OFFSET_LAST_STS_1 = 0x104, > + PVSCSI_REG_OFFSET_LAST_STS_2 = 0x108, > + PVSCSI_REG_OFFSET_LAST_STS_3 = 0x10c, > + PVSCSI_REG_OFFSET_INTR_STATUS = 0x100c, > + PVSCSI_REG_OFFSET_INTR_MASK = 0x2010, > + PVSCSI_REG_OFFSET_KICK_NON_RW_IO = 0x3014, > + PVSCSI_REG_OFFSET_DEBUG = 0x3018, > + PVSCSI_REG_OFFSET_KICK_RW_IO = 0x4018, > +}; > + > +enum PVSCSICommands { > + PVSCSI_CMD_FIRST = 0, > + PVSCSI_CMD_ADAPTER_RESET = 1, > + PVSCSI_CMD_ISSUE_SCSI = 2, > + PVSCSI_CMD_SETUP_RINGS = 3, > + PVSCSI_CMD_RESET_BUS = 4, > + PVSCSI_CMD_RESET_DEVICE = 5, > + PVSCSI_CMD_ABORT_CMD = 6, > + PVSCSI_CMD_CONFIG = 7, > + PVSCSI_CMD_SETUP_MSG_RING = 8, > + PVSCSI_CMD_DEVICE_UNPLUG = 9, > + PVSCSI_CMD_LAST = 10 > +}; > + > +#define PVSCSI_SETUP_RINGS_MAX_NUM_PAGES 32 > +struct PVSCSICmdDescSetupRings { > + u32 reqRingNumPages; > + u32 cmpRingNumPages; > + u64 ringsStatePPN; > + u64 reqRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES]; > + u64 cmpRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES]; > +} PACKED; > + > +struct PVSCSIRingCmpDesc { > + u64 context; > + u64 dataLen; > + u32 senseLen; > + u16 hostStatus; > + u16 scsiStatus; > + u32 pad[2]; > +} PACKED; > + > +struct PVSCSIRingsState { > + u32 reqProdIdx; > + u32 reqConsIdx; > + u32 reqNumEntriesLog2; > + > + u32 cmpProdIdx; > + u32 cmpConsIdx; > + u32 cmpNumEntriesLog2; > + > + u8 pad[104]; > + > + u32 msgProdIdx; > + u32 msgConsIdx; > + u32 msgNumEntriesLog2; > +} PACKED; > + > +struct PVSCSIRingReqDesc { > + u64 context; > + u64 dataAddr; > + u64 dataLen; > + u64 senseAddr; > + u32 senseLen; > + u32 flags; > + u8 cdb[16]; > + u8 cdbLen; > + u8 lun[8]; > + u8 tag; > + u8 bus; > + u8 target; > + u8 vcpuHint; > + u8 unused[59]; > +} PACKED; > + > +struct pvscsi_ring_dsc_s { > + struct PVSCSIRingsState *ring_state; > + struct PVSCSIRingReqDesc *ring_reqs; > + struct PVSCSIRingCmpDesc *ring_cmps; > +}; > + > +struct pvscsi_lun_s { > + struct drive_s drive; > + struct pci_device *pci; > + u32 iobase; > + u8 target; > + u8 lun; > + struct pvscsi_ring_dsc_s *ring_dsc; > +}; > + > +static void > +pvscsi_write_cmd_desc(u32 iobase, u32 cmd, const void *desc, size_t len) > +{ > + const u32 *ptr = desc; > + size_t i; > + > + len /= sizeof(*ptr); > + pci_writel(iobase + PVSCSI_REG_OFFSET_COMMAND, cmd); > + for (i = 0; i < len; i++) > + pci_writel(iobase + PVSCSI_REG_OFFSET_COMMAND_DATA, ptr[i]); > +} > + > +static void > +pvscsi_kick_rw_io(u32 iobase) > +{ > + pci_writel(iobase + PVSCSI_REG_OFFSET_KICK_RW_IO, 0); > +} > + > +static void > +pvscsi_wait_intr_cmpl(u32 iobase) > +{ > + while (!(pci_readl(iobase + PVSCSI_REG_OFFSET_INTR_STATUS) & PVSCSI_INTR_CMPL_MASK)) > + usleep(5); > + pci_writel(iobase + PVSCSI_REG_OFFSET_INTR_STATUS, PVSCSI_INTR_CMPL_MASK); > + > +} > + > +static void > +pvscsi_init_rings(u32 iobase, struct pvscsi_ring_dsc_s **ring_dsc) > +{ > + struct PVSCSICmdDescSetupRings cmd = {0,}; > + > + struct pvscsi_ring_dsc_s *dsc = memalign_low(sizeof(*dsc), PAGE_SIZE); > + if (!dsc) { > + warn_noalloc(); > + return; > + } > + > + dsc->ring_state = > + (struct PVSCSIRingsState *)memalign_low(PAGE_SIZE, PAGE_SIZE); > + dsc->ring_reqs = > + (struct PVSCSIRingReqDesc *)memalign_low(PAGE_SIZE, PAGE_SIZE); > + dsc->ring_cmps = > + (struct PVSCSIRingCmpDesc *)memalign_low(PAGE_SIZE, PAGE_SIZE); > + if (!dsc->ring_state || !dsc->ring_reqs || !dsc->ring_cmps) { > + warn_noalloc(); > + return; > + } > + memset(dsc->ring_state, 0, PAGE_SIZE); > + memset(dsc->ring_reqs, 0, PAGE_SIZE); > + memset(dsc->ring_cmps, 0, PAGE_SIZE); > + > + cmd.reqRingNumPages = 1; > + cmd.cmpRingNumPages = 1; > + cmd.ringsStatePPN = virt_to_phys(dsc->ring_state) >> PAGE_SHIFT; > + cmd.reqRingPPNs[0] = virt_to_phys(dsc->ring_reqs) >> PAGE_SHIFT; > + cmd.cmpRingPPNs[0] = virt_to_phys(dsc->ring_cmps) >> PAGE_SHIFT; > + > + pvscsi_write_cmd_desc(iobase, PVSCSI_CMD_SETUP_RINGS, > + &cmd, sizeof(cmd)); > + *ring_dsc = dsc; > +} > + > +static void pvscsi_fill_req(struct PVSCSIRingsState *s, > + struct PVSCSIRingReqDesc *req, > + u16 target, u16 lun, void *cdbcmd, u16 blocksize, > + struct disk_op_s *op) > +{ > + SET_LOWFLAT(req->bus, 0); > + SET_LOWFLAT(req->target, target); > + memset(LOWFLAT2LOW(&req->lun[0]), 0, sizeof(req->lun)); > + SET_LOWFLAT(req->lun[1], lun); > + SET_LOWFLAT(req->senseLen, 0); > + SET_LOWFLAT(req->senseAddr, 0); > + SET_LOWFLAT(req->cdbLen, 16); > + SET_LOWFLAT(req->vcpuHint, 0); > + memcpy(LOWFLAT2LOW(&req->cdb[0]), cdbcmd, 16); > + SET_LOWFLAT(req->tag, SIMPLE_QUEUE_TAG); > + SET_LOWFLAT(req->flags, > + cdb_is_read(cdbcmd, blocksize) ? > + PVSCSI_FLAG_CMD_DIR_TOHOST : PVSCSI_FLAG_CMD_DIR_TODEVICE); > + > + SET_LOWFLAT(req->dataLen, op->count * blocksize); > + SET_LOWFLAT(req->dataAddr, (u32)op->buf_fl); > + SET_LOWFLAT(s->reqProdIdx, GET_LOWFLAT(s->reqProdIdx) + 1); > + > +} > + > +static u32 > +pvscsi_get_rsp(struct PVSCSIRingsState *s, > + struct PVSCSIRingCmpDesc *rsp) > +{ > + u32 status = GET_LOWFLAT(rsp->hostStatus); > + SET_LOWFLAT(s->cmpConsIdx, GET_LOWFLAT(s->cmpConsIdx)+1); > + return status; > +} > + > +static int > +pvscsi_cmd(struct pvscsi_lun_s *plun, struct disk_op_s *op, > + void *cdbcmd, u16 target, u16 lun, u16 blocksize) > +{ > + struct pvscsi_ring_dsc_s *ring_dsc = GET_GLOBAL(plun->ring_dsc); > + struct PVSCSIRingsState *s = GET_LOWFLAT(ring_dsc->ring_state); > + u32 req_entries = GET_LOWFLAT(s->reqNumEntriesLog2); > + u32 cmp_entries = GET_LOWFLAT(s->cmpNumEntriesLog2); > + struct PVSCSIRingReqDesc *req; > + struct PVSCSIRingCmpDesc *rsp; > + u32 status; > + > + if (GET_LOWFLAT(s->reqProdIdx) - GET_LOWFLAT(s->cmpConsIdx) >= 1 << req_entries) { > + dprintf(1, "pvscsi: ring full: reqProdIdx=%d cmpConsIdx=%d\n", > + GET_LOWFLAT(s->reqProdIdx), GET_LOWFLAT(s->cmpConsIdx)); > + return DISK_RET_EBADTRACK; > + } > + > + req = GET_LOWFLAT(ring_dsc->ring_reqs) + (GET_LOWFLAT(s->reqProdIdx) & MASK(req_entries)); > + pvscsi_fill_req(s, req, target, lun, cdbcmd, blocksize, op); > + > + pvscsi_kick_rw_io(GET_GLOBAL(plun->iobase)); > + pvscsi_wait_intr_cmpl(GET_GLOBAL(plun->iobase)); > + > + rsp = GET_LOWFLAT(ring_dsc->ring_cmps) + (GET_LOWFLAT(s->cmpConsIdx) & MASK(cmp_entries)); > + status = pvscsi_get_rsp(s, rsp); > + > + return status == 0 ? DISK_RET_SUCCESS : DISK_RET_EBADTRACK; > +} > + > +int > +pvscsi_cmd_data(struct disk_op_s *op, void *cdbcmd, u16 blocksize) > +{ > + if (!CONFIG_PVSCSI) > + return DISK_RET_EBADTRACK; > + > + struct pvscsi_lun_s *plun = > + container_of(op->drive_g, struct pvscsi_lun_s, drive); > + > + return pvscsi_cmd(plun, op, cdbcmd, > + GET_GLOBAL(plun->target), GET_GLOBAL(plun->lun), > + blocksize); > + > +} > + > +static int > +pvscsi_add_lun(struct pci_device *pci, u32 iobase, > + struct pvscsi_ring_dsc_s *ring_dsc, u8 target, u8 lun) > +{ > + struct pvscsi_lun_s *plun = malloc_fseg(sizeof(*plun)); > + if (!plun) { > + warn_noalloc(); > + return -1; > + } > + memset(plun, 0, sizeof(*plun)); I missed that yesterday. This `memset()` seems unneeded too as below all members are set explicitly. > + plun->drive.type = DTYPE_PVSCSI; > + plun->drive.cntl_id = pci->bdf; > + plun->pci = pci; > + plun->target = target; > + plun->lun = lun; > + plun->iobase = iobase; > + plun->ring_dsc = ring_dsc; […] Kevin, could you please reply, if another iteration is needed or if you can remove that `memset()` yourself, when applying this patch? Acked-by: Paul Menzel <paulepanter(a)users.sourceforge.net> Thanks, Paul

2 1

[PATCH v4] Add pvscsi boot support
by Evgeny Budilovsky 14 Oct '13

14 Oct '13

Testing was done on windows images (win 2008/2012) taken from esx with vmware tools installed and boot disk configured to use pvscsi. Also I've used linux (ubuntu 12.04) where pvscsi drivers are installed by default and booted it using qemu cmd similar to this: ./x86_64-softmmu/qemu-system-x86_64 -snapshot -device pvscsi,id=pvscsi0 \ -device scsi-disk,bus=pvscsi0.0,drive=drive0 \ -drive id=drive0,if=none,file=ubuntu-12.04.qcow2 \ -bios roms/seabios/out/bios.bin Signed-off-by: Evgeny Budilovsky <evgeny.budilovsky(a)ravellosystems.com> --- Makefile | 2 +- src/Kconfig | 12 ++ src/block.c | 1 + src/block.h | 1 + src/hw/blockcmd.c | 3 + src/hw/pci_ids.h | 3 + src/hw/pvscsi.c | 363 +++++++++++++++++++++++++++++++++++++++++++++++++++++ src/hw/pvscsi.h | 8 ++ src/post.c | 2 + 9 files changed, 394 insertions(+), 1 deletion(-) create mode 100644 src/hw/pvscsi.c create mode 100644 src/hw/pvscsi.h diff --git a/Makefile b/Makefile index 3218970..6c6302e 100644 --- a/Makefile +++ b/Makefile @@ -34,7 +34,7 @@ SRCBOTH=misc.c stacks.c output.c string.c x86.c block.c cdrom.c mouse.c kbd.c \ hw/usb-hid.c hw/usb-msc.c hw/usb-uas.c \ hw/blockcmd.c hw/floppy.c hw/ata.c hw/ahci.c hw/ramdisk.c \ hw/virtio-ring.c hw/virtio-pci.c hw/virtio-blk.c hw/virtio-scsi.c \ - hw/lsi-scsi.c hw/esp-scsi.c hw/megasas.c + hw/lsi-scsi.c hw/esp-scsi.c hw/megasas.c hw/pvscsi.c SRC16=$(SRCBOTH) system.c disk.c font.c SRC32FLAT=$(SRCBOTH) post.c memmap.c malloc.c pmm.c romfile.c optionroms.c \ boot.c bootsplash.c jpeg.c bmp.c \ diff --git a/src/Kconfig b/src/Kconfig index c40cc61..38e24bf 100644 --- a/src/Kconfig +++ b/src/Kconfig @@ -161,6 +161,18 @@ menu "Hardware support" default y help Support boot from virtio-scsi storage. + config PVSCSI + depends on DRIVES && QEMU_HARDWARE + bool "PVSCSI controllers" + default y + help + Support boot from Paravirtualized SCSI storage. This kind of storage + is mainly supported by VMware ESX hypervisor. It is commonly used + to allow fast storage access by communicating directly with the + underlying hypervisor. Enabling this type of boot will allow + booting directly from images imported from an ESX platform, + without the need to use slower emulation of storage controllers + such as IDE. config ESP_SCSI depends on DRIVES && QEMU_HARDWARE bool "AMD PCscsi controllers" diff --git a/src/block.c b/src/block.c index cbcf9e1..2d01f6a 100644 --- a/src/block.c +++ b/src/block.c @@ -373,6 +373,7 @@ process_op(struct disk_op_s *op) case DTYPE_LSI_SCSI: case DTYPE_ESP_SCSI: case DTYPE_MEGASAS: + case DTYPE_PVSCSI: return process_scsi_op(op); default: op->count = 0; diff --git a/src/block.h b/src/block.h index 3492806..aea6a71 100644 --- a/src/block.h +++ b/src/block.h @@ -83,6 +83,7 @@ struct drive_s { #define DTYPE_LSI_SCSI 0x0c #define DTYPE_ESP_SCSI 0x0d #define DTYPE_MEGASAS 0x0e +#define DTYPE_PVSCSI 0x0f #define MAXDESCSIZE 80 diff --git a/src/hw/blockcmd.c b/src/hw/blockcmd.c index 7bdcf38..39d3caa 100644 --- a/src/hw/blockcmd.c +++ b/src/hw/blockcmd.c @@ -14,6 +14,7 @@ #include "esp-scsi.h" // esp_scsi_cmd_data #include "lsi-scsi.h" // lsi_scsi_cmd_data #include "megasas.h" // megasas_cmd_data +#include "pvscsi.h" // pvscsi_cmd_data #include "output.h" // dprintf #include "std/disk.h" // DISK_RET_EPARAM #include "string.h" // memset @@ -44,6 +45,8 @@ cdb_cmd_data(struct disk_op_s *op, void *cdbcmd, u16 blocksize) return esp_scsi_cmd_data(op, cdbcmd, blocksize); case DTYPE_MEGASAS: return megasas_cmd_data(op, cdbcmd, blocksize); + case DTYPE_PVSCSI: + return pvscsi_cmd_data(op, cdbcmd, blocksize); default: op->count = 0; return DISK_RET_EPARAM; diff --git a/src/hw/pci_ids.h b/src/hw/pci_ids.h index 322d156..1cd4f72 100644 --- a/src/hw/pci_ids.h +++ b/src/hw/pci_ids.h @@ -2618,3 +2618,6 @@ #define PCI_VENDOR_ID_REDHAT_QUMRANET 0x1af4 #define PCI_DEVICE_ID_VIRTIO_BLK 0x1001 #define PCI_DEVICE_ID_VIRTIO_SCSI 0x1004 + +#define PCI_VENDOR_ID_VMWARE 0x15ad +#define PCI_DEVICE_ID_VMWARE_PVSCSI 0x07C0 diff --git a/src/hw/pvscsi.c b/src/hw/pvscsi.c new file mode 100644 index 0000000..740a74b --- /dev/null +++ b/src/hw/pvscsi.c @@ -0,0 +1,363 @@ +// QEMU VMWARE Paravirtualized SCSI boot support. +// +// Copyright (c) 2013 Ravello Systems LTD (http://ravellosystems.com) +// +// Authors: +// Evgeny Budilovsky <evgeny.budilovsky(a)ravellosystems.com> +// +// This file may be distributed under the terms of the GNU LGPLv3 license. + +#include "biosvar.h" // GET_GLOBAL +#include "block.h" // struct drive_s +#include "blockcmd.h" // scsi_drive_setup +#include "config.h" // CONFIG_* +#include "malloc.h" // free +#include "output.h" // dprintf +#include "pci.h" // foreachpci +#include "pci_ids.h" // PCI_DEVICE_ID_VMWARE_PVSCSI +#include "pci_regs.h" // PCI_VENDOR_ID +#include "std/disk.h" // DISK_RET_SUCCESS +#include "string.h" // memset +#include "util.h" // usleep +#include "pvscsi.h" +#include "virtio-ring.h" // PAGE_SHIFT, virt_to_phys + +#define MASK(n) ((1 << (n)) - 1) + +#define SIMPLE_QUEUE_TAG 0x20 + +#define PVSCSI_INTR_CMPL_0 (1 << 0) +#define PVSCSI_INTR_CMPL_1 (1 << 1) +#define PVSCSI_INTR_CMPL_MASK MASK(2) + +#define PVSCSI_INTR_MSG_0 (1 << 2) +#define PVSCSI_INTR_MSG_1 (1 << 3) +#define PVSCSI_INTR_MSG_MASK (MASK(2) << 2) +#define PVSCSI_INTR_ALL_SUPPORTED MASK(4) + +#define PVSCSI_FLAG_CMD_WITH_SG_LIST (1 << 0) +#define PVSCSI_FLAG_CMD_OUT_OF_BAND_CDB (1 << 1) +#define PVSCSI_FLAG_CMD_DIR_NONE (1 << 2) +#define PVSCSI_FLAG_CMD_DIR_TOHOST (1 << 3) +#define PVSCSI_FLAG_CMD_DIR_TODEVICE (1 << 4) + +enum PVSCSIRegOffset { + PVSCSI_REG_OFFSET_COMMAND = 0x0, + PVSCSI_REG_OFFSET_COMMAND_DATA = 0x4, + PVSCSI_REG_OFFSET_COMMAND_STATUS = 0x8, + PVSCSI_REG_OFFSET_LAST_STS_0 = 0x100, + PVSCSI_REG_OFFSET_LAST_STS_1 = 0x104, + PVSCSI_REG_OFFSET_LAST_STS_2 = 0x108, + PVSCSI_REG_OFFSET_LAST_STS_3 = 0x10c, + PVSCSI_REG_OFFSET_INTR_STATUS = 0x100c, + PVSCSI_REG_OFFSET_INTR_MASK = 0x2010, + PVSCSI_REG_OFFSET_KICK_NON_RW_IO = 0x3014, + PVSCSI_REG_OFFSET_DEBUG = 0x3018, + PVSCSI_REG_OFFSET_KICK_RW_IO = 0x4018, +}; + +enum PVSCSICommands { + PVSCSI_CMD_FIRST = 0, + PVSCSI_CMD_ADAPTER_RESET = 1, + PVSCSI_CMD_ISSUE_SCSI = 2, + PVSCSI_CMD_SETUP_RINGS = 3, + PVSCSI_CMD_RESET_BUS = 4, + PVSCSI_CMD_RESET_DEVICE = 5, + PVSCSI_CMD_ABORT_CMD = 6, + PVSCSI_CMD_CONFIG = 7, + PVSCSI_CMD_SETUP_MSG_RING = 8, + PVSCSI_CMD_DEVICE_UNPLUG = 9, + PVSCSI_CMD_LAST = 10 +}; + +#define PVSCSI_SETUP_RINGS_MAX_NUM_PAGES 32 +struct PVSCSICmdDescSetupRings { + u32 reqRingNumPages; + u32 cmpRingNumPages; + u64 ringsStatePPN; + u64 reqRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES]; + u64 cmpRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES]; +} PACKED; + +struct PVSCSIRingCmpDesc { + u64 context; + u64 dataLen; + u32 senseLen; + u16 hostStatus; + u16 scsiStatus; + u32 pad[2]; +} PACKED; + +struct PVSCSIRingsState { + u32 reqProdIdx; + u32 reqConsIdx; + u32 reqNumEntriesLog2; + + u32 cmpProdIdx; + u32 cmpConsIdx; + u32 cmpNumEntriesLog2; + + u8 pad[104]; + + u32 msgProdIdx; + u32 msgConsIdx; + u32 msgNumEntriesLog2; +} PACKED; + +struct PVSCSIRingReqDesc { + u64 context; + u64 dataAddr; + u64 dataLen; + u64 senseAddr; + u32 senseLen; + u32 flags; + u8 cdb[16]; + u8 cdbLen; + u8 lun[8]; + u8 tag; + u8 bus; + u8 target; + u8 vcpuHint; + u8 unused[59]; +} PACKED; + +struct pvscsi_ring_dsc_s { + struct PVSCSIRingsState *ring_state; + struct PVSCSIRingReqDesc *ring_reqs; + struct PVSCSIRingCmpDesc *ring_cmps; +}; + +struct pvscsi_lun_s { + struct drive_s drive; + struct pci_device *pci; + u32 iobase; + u8 target; + u8 lun; + struct pvscsi_ring_dsc_s *ring_dsc; +}; + +static void +pvscsi_write_cmd_desc(u32 iobase, u32 cmd, const void *desc, size_t len) +{ + const u32 *ptr = desc; + size_t i; + + len /= sizeof(*ptr); + pci_writel(iobase + PVSCSI_REG_OFFSET_COMMAND, cmd); + for (i = 0; i < len; i++) + pci_writel(iobase + PVSCSI_REG_OFFSET_COMMAND_DATA, ptr[i]); +} + +static void +pvscsi_kick_rw_io(u32 iobase) +{ + pci_writel(iobase + PVSCSI_REG_OFFSET_KICK_RW_IO, 0); +} + +static void +pvscsi_wait_intr_cmpl(u32 iobase) +{ + while (!(pci_readl(iobase + PVSCSI_REG_OFFSET_INTR_STATUS) & PVSCSI_INTR_CMPL_MASK)) + usleep(5); + pci_writel(iobase + PVSCSI_REG_OFFSET_INTR_STATUS, PVSCSI_INTR_CMPL_MASK); + +} + +static void +pvscsi_init_rings(u32 iobase, struct pvscsi_ring_dsc_s **ring_dsc) +{ + struct PVSCSICmdDescSetupRings cmd = {0,}; + + struct pvscsi_ring_dsc_s *dsc = memalign_low(sizeof(*dsc), PAGE_SIZE); + if (!dsc) { + warn_noalloc(); + return; + } + + dsc->ring_state = + (struct PVSCSIRingsState *)memalign_low(PAGE_SIZE, PAGE_SIZE); + dsc->ring_reqs = + (struct PVSCSIRingReqDesc *)memalign_low(PAGE_SIZE, PAGE_SIZE); + dsc->ring_cmps = + (struct PVSCSIRingCmpDesc *)memalign_low(PAGE_SIZE, PAGE_SIZE); + if (!dsc->ring_state || !dsc->ring_reqs || !dsc->ring_cmps) { + warn_noalloc(); + return; + } + memset(dsc->ring_state, 0, PAGE_SIZE); + memset(dsc->ring_reqs, 0, PAGE_SIZE); + memset(dsc->ring_cmps, 0, PAGE_SIZE); + + cmd.reqRingNumPages = 1; + cmd.cmpRingNumPages = 1; + cmd.ringsStatePPN = virt_to_phys(dsc->ring_state) >> PAGE_SHIFT; + cmd.reqRingPPNs[0] = virt_to_phys(dsc->ring_reqs) >> PAGE_SHIFT; + cmd.cmpRingPPNs[0] = virt_to_phys(dsc->ring_cmps) >> PAGE_SHIFT; + + pvscsi_write_cmd_desc(iobase, PVSCSI_CMD_SETUP_RINGS, + &cmd, sizeof(cmd)); + *ring_dsc = dsc; +} + +static void pvscsi_fill_req(struct PVSCSIRingsState *s, + struct PVSCSIRingReqDesc *req, + u16 target, u16 lun, void *cdbcmd, u16 blocksize, + struct disk_op_s *op) +{ + SET_LOWFLAT(req->bus, 0); + SET_LOWFLAT(req->target, target); + memset(LOWFLAT2LOW(&req->lun[0]), 0, sizeof(req->lun)); + SET_LOWFLAT(req->lun[1], lun); + SET_LOWFLAT(req->senseLen, 0); + SET_LOWFLAT(req->senseAddr, 0); + SET_LOWFLAT(req->cdbLen, 16); + SET_LOWFLAT(req->vcpuHint, 0); + memcpy(LOWFLAT2LOW(&req->cdb[0]), cdbcmd, 16); + SET_LOWFLAT(req->tag, SIMPLE_QUEUE_TAG); + SET_LOWFLAT(req->flags, + cdb_is_read(cdbcmd, blocksize) ? + PVSCSI_FLAG_CMD_DIR_TOHOST : PVSCSI_FLAG_CMD_DIR_TODEVICE); + + SET_LOWFLAT(req->dataLen, op->count * blocksize); + SET_LOWFLAT(req->dataAddr, (u32)op->buf_fl); + SET_LOWFLAT(s->reqProdIdx, GET_LOWFLAT(s->reqProdIdx) + 1); + +} + +static u32 +pvscsi_get_rsp(struct PVSCSIRingsState *s, + struct PVSCSIRingCmpDesc *rsp) +{ + u32 status = GET_LOWFLAT(rsp->hostStatus); + SET_LOWFLAT(s->cmpConsIdx, GET_LOWFLAT(s->cmpConsIdx)+1); + return status; +} + +static int +pvscsi_cmd(struct pvscsi_lun_s *plun, struct disk_op_s *op, + void *cdbcmd, u16 target, u16 lun, u16 blocksize) +{ + struct pvscsi_ring_dsc_s *ring_dsc = GET_GLOBAL(plun->ring_dsc); + struct PVSCSIRingsState *s = GET_LOWFLAT(ring_dsc->ring_state); + u32 req_entries = GET_LOWFLAT(s->reqNumEntriesLog2); + u32 cmp_entries = GET_LOWFLAT(s->cmpNumEntriesLog2); + struct PVSCSIRingReqDesc *req; + struct PVSCSIRingCmpDesc *rsp; + u32 status; + + if (GET_LOWFLAT(s->reqProdIdx) - GET_LOWFLAT(s->cmpConsIdx) >= 1 << req_entries) { + dprintf(1, "pvscsi: ring full: reqProdIdx=%d cmpConsIdx=%d\n", + GET_LOWFLAT(s->reqProdIdx), GET_LOWFLAT(s->cmpConsIdx)); + return DISK_RET_EBADTRACK; + } + + req = GET_LOWFLAT(ring_dsc->ring_reqs) + (GET_LOWFLAT(s->reqProdIdx) & MASK(req_entries)); + pvscsi_fill_req(s, req, target, lun, cdbcmd, blocksize, op); + + pvscsi_kick_rw_io(GET_GLOBAL(plun->iobase)); + pvscsi_wait_intr_cmpl(GET_GLOBAL(plun->iobase)); + + rsp = GET_LOWFLAT(ring_dsc->ring_cmps) + (GET_LOWFLAT(s->cmpConsIdx) & MASK(cmp_entries)); + status = pvscsi_get_rsp(s, rsp); + + return status == 0 ? DISK_RET_SUCCESS : DISK_RET_EBADTRACK; +} + +int +pvscsi_cmd_data(struct disk_op_s *op, void *cdbcmd, u16 blocksize) +{ + if (!CONFIG_PVSCSI) + return DISK_RET_EBADTRACK; + + struct pvscsi_lun_s *plun = + container_of(op->drive_g, struct pvscsi_lun_s, drive); + + return pvscsi_cmd(plun, op, cdbcmd, + GET_GLOBAL(plun->target), GET_GLOBAL(plun->lun), + blocksize); + +} + +static int +pvscsi_add_lun(struct pci_device *pci, u32 iobase, + struct pvscsi_ring_dsc_s *ring_dsc, u8 target, u8 lun) +{ + struct pvscsi_lun_s *plun = malloc_fseg(sizeof(*plun)); + if (!plun) { + warn_noalloc(); + return -1; + } + memset(plun, 0, sizeof(*plun)); + plun->drive.type = DTYPE_PVSCSI; + plun->drive.cntl_id = pci->bdf; + plun->pci = pci; + plun->target = target; + plun->lun = lun; + plun->iobase = iobase; + plun->ring_dsc = ring_dsc; + + char *name = znprintf(16, "pvscsi %02x:%02x.%x %d:%d", + pci_bdf_to_bus(pci->bdf), pci_bdf_to_dev(pci->bdf), + pci_bdf_to_fn(pci->bdf), target, lun); + int prio = bootprio_find_scsi_device(pci, target, lun); + int ret = scsi_drive_setup(&plun->drive, name, prio); + free(name); + if (ret) + goto fail; + return 0; + +fail: + free(plun); + return -1; +} + +static void +pvscsi_scan_target(struct pci_device *pci, u32 iobase, + struct pvscsi_ring_dsc_s *ring_dsc, u8 target) +{ + /* TODO: send REPORT LUNS. For now, only LUN 0 is recognized. */ + pvscsi_add_lun(pci, iobase, ring_dsc, target, 0); +} + +static void +init_pvscsi(struct pci_device *pci) +{ + struct pvscsi_ring_dsc_s *ring_dsc = NULL; + int i; + u16 bdf = pci->bdf; + u32 iobase = pci_config_readl(pci->bdf, PCI_BASE_ADDRESS_0) + & PCI_BASE_ADDRESS_MEM_MASK; + + pci_config_maskw(bdf, PCI_COMMAND, 0, PCI_COMMAND_MASTER); + + dprintf(1, "found pvscsi at %02x:%02x.%x, io @ %x\n", + pci_bdf_to_bus(bdf), pci_bdf_to_dev(bdf), + pci_bdf_to_fn(bdf), iobase); + + pvscsi_write_cmd_desc(iobase, PVSCSI_CMD_ADAPTER_RESET, NULL, 0); + + pvscsi_init_rings(iobase, &ring_dsc); + for (i = 0; i < 7; i++) + pvscsi_scan_target(pci, iobase, ring_dsc, i); + + return; + +} + +void +pvscsi_setup(void) +{ + ASSERT32FLAT(); + if (! CONFIG_PVSCSI) + return; + + dprintf(3, "init pvscsi\n"); + + struct pci_device *pci; + foreachpci(pci) { + if (pci->vendor != PCI_VENDOR_ID_VMWARE + || pci->device != PCI_DEVICE_ID_VMWARE_PVSCSI) + continue; + init_pvscsi(pci); + } +} diff --git a/src/hw/pvscsi.h b/src/hw/pvscsi.h new file mode 100644 index 0000000..fde9f0b --- /dev/null +++ b/src/hw/pvscsi.h @@ -0,0 +1,8 @@ +#ifndef _PVSCSI_H_ +#define _PVSCSI_H_ + +struct disk_op_s; +int pvscsi_cmd_data(struct disk_op_s *op, void *cdbcmd, u16 blocksize); +void pvscsi_setup(void); + +#endif /* _PVSCSI_H_ */ diff --git a/src/post.c b/src/post.c index acd2362..b634686 100644 --- a/src/post.c +++ b/src/post.c @@ -15,6 +15,7 @@ #include "hw/esp-scsi.h" // esp_scsi_setup #include "hw/lsi-scsi.h" // lsi_scsi_setup #include "hw/megasas.h" // megasas_setup +#include "hw/pvscsi.h" // pvscsi_setup #include "hw/pic.h" // pic_setup #include "hw/ps2port.h" // ps2port_setup #include "hw/rtc.h" // rtc_write @@ -147,6 +148,7 @@ device_hardware_setup(void) lsi_scsi_setup(); esp_scsi_setup(); megasas_setup(); + pvscsi_setup(); } static void -- 1.7.9.5

1 0

Re: [SeaBIOS] [PATCH v3] pvscsi boot support
by Paul Menzel 14 Oct '13

14 Oct '13

Dear Evgeny, thank you for the patch iterations! A minor nitpick, I prefer having a verb in the commit summary. Add pvscsi boot support Also you could add your reply (the things you tested on) to the commit message. Am Sonntag, den 13.10.2013, 20:07 +0300 schrieb Evgeny Budilovsky: > Signed-off-by: Evgeny Budilovsky <evgeny.budilovsky(a)ravellosystems.com> > --- > Makefile | 2 +- > src/Kconfig | 12 ++ > src/block.c | 1 + > src/block.h | 1 + > src/hw/blockcmd.c | 3 + > src/hw/pci_ids.h | 3 + > src/hw/pvscsi.c | 365 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > src/hw/pvscsi.h | 8 ++ > src/post.c | 2 + > 9 files changed, 396 insertions(+), 1 deletion(-) > create mode 100644 src/hw/pvscsi.c > create mode 100644 src/hw/pvscsi.h > > diff --git a/Makefile b/Makefile > index 3218970..6c6302e 100644 > --- a/Makefile > +++ b/Makefile > @@ -34,7 +34,7 @@ SRCBOTH=misc.c stacks.c output.c string.c x86.c block.c cdrom.c mouse.c kbd.c \ > hw/usb-hid.c hw/usb-msc.c hw/usb-uas.c \ > hw/blockcmd.c hw/floppy.c hw/ata.c hw/ahci.c hw/ramdisk.c \ > hw/virtio-ring.c hw/virtio-pci.c hw/virtio-blk.c hw/virtio-scsi.c \ > - hw/lsi-scsi.c hw/esp-scsi.c hw/megasas.c > + hw/lsi-scsi.c hw/esp-scsi.c hw/megasas.c hw/pvscsi.c > SRC16=$(SRCBOTH) system.c disk.c font.c > SRC32FLAT=$(SRCBOTH) post.c memmap.c malloc.c pmm.c romfile.c optionroms.c \ > boot.c bootsplash.c jpeg.c bmp.c \ > diff --git a/src/Kconfig b/src/Kconfig > index c40cc61..6e6814e 100644 > --- a/src/Kconfig > +++ b/src/Kconfig > @@ -161,6 +161,18 @@ menu "Hardware support" > default y > help > Support boot from virtio-scsi storage. > + config PVSCSI > + depends on DRIVES && QEMU_HARDWARE > + bool "pvscsi controllers" Should it be capitalizied? > + default y > + help > + Support boot from Paravirtualized SCSI storage. This kind of storage > + is mainly supported by VMware ESX hypervisor. It is commonly used > + to allow fast storage access by communicating directly with the > + underlying hypervisor. Enabling this type of boot will allow > + booting directly from images that were imported from ESX platform, »ESX platform*s*« or »from an ESX platform«? (I am no native speaker though.) To make it shorter remove the »that were«: … imported from images imported from … (as above). > + without the need to use slower emulation of storage controllers > + such as IDE. […] > diff --git a/src/hw/pvscsi.c b/src/hw/pvscsi.c > new file mode 100644 > index 0000000..df5faeb > --- /dev/null > +++ b/src/hw/pvscsi.c > @@ -0,0 +1,365 @@ > +// QEMU VMWARE Paravirtualized SCSI boot support. > +// > +// Copyright (c) 2013 Ravello Systems LTD (http://ravellosystems.com) > +// > +// Authors: > +// Evgeny Budilovsky <evgeny.budilovsky(a)ravellosystems.com> > +// > +// This file may be distributed under the terms of the GNU LGPLv3 license. > + > +#include "biosvar.h" // GET_GLOBAL > +#include "block.h" // struct drive_s > +#include "blockcmd.h" // scsi_drive_setup > +#include "config.h" // CONFIG_* > +#include "malloc.h" // free > +#include "output.h" // dprintf > +#include "pci.h" // foreachpci > +#include "pci_ids.h" // PCI_DEVICE_ID_VMWARE_PVSCSI > +#include "pci_regs.h" // PCI_VENDOR_ID > +#include "std/disk.h" // DISK_RET_SUCCESS > +#include "string.h" // memset > +#include "util.h" // usleep > +#include "pvscsi.h" > +#include "virtio-ring.h" // PAGE_SHIFT, virt_to_phys > + > +#define MASK(n) ((1 << (n)) - 1) > + > +#define SIMPLE_QUEUE_TAG 0x20 > + > +#define PVSCSI_INTR_CMPL_0 (1 << 0) > +#define PVSCSI_INTR_CMPL_1 (1 << 1) > +#define PVSCSI_INTR_CMPL_MASK MASK(2) > + > +#define PVSCSI_INTR_MSG_0 (1 << 2) > +#define PVSCSI_INTR_MSG_1 (1 << 3) > +#define PVSCSI_INTR_MSG_MASK (MASK(2) << 2) > +#define PVSCSI_INTR_ALL_SUPPORTED MASK(4) > + > +#define PVSCSI_FLAG_CMD_WITH_SG_LIST (1 << 0) > +#define PVSCSI_FLAG_CMD_OUT_OF_BAND_CDB (1 << 1) > +#define PVSCSI_FLAG_CMD_DIR_NONE (1 << 2) > +#define PVSCSI_FLAG_CMD_DIR_TOHOST (1 << 3) > +#define PVSCSI_FLAG_CMD_DIR_TODEVICE (1 << 4) > + > +enum PVSCSIRegOffset { > + PVSCSI_REG_OFFSET_COMMAND = 0x0, > + PVSCSI_REG_OFFSET_COMMAND_DATA = 0x4, > + PVSCSI_REG_OFFSET_COMMAND_STATUS = 0x8, > + PVSCSI_REG_OFFSET_LAST_STS_0 = 0x100, > + PVSCSI_REG_OFFSET_LAST_STS_1 = 0x104, > + PVSCSI_REG_OFFSET_LAST_STS_2 = 0x108, > + PVSCSI_REG_OFFSET_LAST_STS_3 = 0x10c, > + PVSCSI_REG_OFFSET_INTR_STATUS = 0x100c, > + PVSCSI_REG_OFFSET_INTR_MASK = 0x2010, > + PVSCSI_REG_OFFSET_KICK_NON_RW_IO = 0x3014, > + PVSCSI_REG_OFFSET_DEBUG = 0x3018, > + PVSCSI_REG_OFFSET_KICK_RW_IO = 0x4018, > +}; > + > +enum PVSCSICommands { > + PVSCSI_CMD_FIRST = 0, > + PVSCSI_CMD_ADAPTER_RESET = 1, > + PVSCSI_CMD_ISSUE_SCSI = 2, > + PVSCSI_CMD_SETUP_RINGS = 3, > + PVSCSI_CMD_RESET_BUS = 4, > + PVSCSI_CMD_RESET_DEVICE = 5, > + PVSCSI_CMD_ABORT_CMD = 6, > + PVSCSI_CMD_CONFIG = 7, > + PVSCSI_CMD_SETUP_MSG_RING = 8, > + PVSCSI_CMD_DEVICE_UNPLUG = 9, > + PVSCSI_CMD_LAST = 10 > +}; > + > +#define PVSCSI_SETUP_RINGS_MAX_NUM_PAGES 32 > +struct PVSCSICmdDescSetupRings { > + u32 reqRingNumPages; > + u32 cmpRingNumPages; > + u64 ringsStatePPN; > + u64 reqRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES]; > + u64 cmpRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES]; > +} PACKED; > + > +struct PVSCSIRingCmpDesc { > + u64 context; > + u64 dataLen; > + u32 senseLen; > + u16 hostStatus; > + u16 scsiStatus; > + u32 pad[2]; > +} PACKED; > + > +struct PVSCSIRingsState { > + u32 reqProdIdx; > + u32 reqConsIdx; > + u32 reqNumEntriesLog2; > + > + u32 cmpProdIdx; > + u32 cmpConsIdx; > + u32 cmpNumEntriesLog2; > + > + u8 pad[104]; > + > + u32 msgProdIdx; > + u32 msgConsIdx; > + u32 msgNumEntriesLog2; > +} PACKED; > + > +struct PVSCSIRingReqDesc { > + u64 context; > + u64 dataAddr; > + u64 dataLen; > + u64 senseAddr; > + u32 senseLen; > + u32 flags; > + u8 cdb[16]; > + u8 cdbLen; > + u8 lun[8]; > + u8 tag; > + u8 bus; > + u8 target; > + u8 vcpuHint; > + u8 unused[59]; > +} PACKED; > + > +struct pvscsi_ring_dsc_s { > + struct PVSCSIRingsState *ring_state; > + struct PVSCSIRingReqDesc *ring_reqs; > + struct PVSCSIRingCmpDesc *ring_cmps; > +}; > + > +struct pvscsi_lun_s { > + struct drive_s drive; > + struct pci_device *pci; > + u32 iobase; > + u8 target; > + u8 lun; > + struct pvscsi_ring_dsc_s *ring_dsc; > +}; Should these go into the header file? > + > +static void > +pvscsi_write_cmd_desc(u32 iobase, u32 cmd, const void *desc, size_t len) > +{ > + const u32 *ptr = desc; > + size_t i; > + > + len /= sizeof(*ptr); > + pci_writel(iobase + PVSCSI_REG_OFFSET_COMMAND, cmd); > + for (i = 0; i < len; i++) > + pci_writel(iobase + PVSCSI_REG_OFFSET_COMMAND_DATA, ptr[i]); > +} > + > +static void > +pvscsi_kick_rw_io(u32 iobase) > +{ > + pci_writel(iobase + PVSCSI_REG_OFFSET_KICK_RW_IO, 0); > +} > + > +static void > +pvscsi_wait_intr_cmpl(u32 iobase) > +{ > + while (!(pci_readl(iobase + PVSCSI_REG_OFFSET_INTR_STATUS) & PVSCSI_INTR_CMPL_MASK)) > + usleep(5); > + pci_writel(iobase + PVSCSI_REG_OFFSET_INTR_STATUS, PVSCSI_INTR_CMPL_MASK); > + > +} > + > +static void > +pvscsi_init_rings(u32 iobase, struct pvscsi_ring_dsc_s **ring_dsc) > +{ > + struct PVSCSICmdDescSetupRings cmd = {0,}; > + > + struct pvscsi_ring_dsc_s *dsc = memalign_low(sizeof(*dsc), PAGE_SIZE); > + if (!dsc) { > + warn_noalloc(); > + return; > + } > + > + memset(dsc, 0, sizeof(*dsc)); That memset is not needed, right? > + > + dsc->ring_state = > + (struct PVSCSIRingsState *)memalign_low(PAGE_SIZE, PAGE_SIZE); > + dsc->ring_reqs = > + (struct PVSCSIRingReqDesc *)memalign_low(PAGE_SIZE, PAGE_SIZE); > + dsc->ring_cmps = > + (struct PVSCSIRingCmpDesc *)memalign_low(PAGE_SIZE, PAGE_SIZE); > + if (!dsc->ring_state || !dsc->ring_reqs || !dsc->ring_cmps) { > + warn_noalloc(); > + return; > + } > + memset(dsc->ring_state, 0, PAGE_SIZE); > + memset(dsc->ring_reqs, 0, PAGE_SIZE); > + memset(dsc->ring_cmps, 0, PAGE_SIZE); > + > + cmd.reqRingNumPages = 1; > + cmd.cmpRingNumPages = 1; > + cmd.ringsStatePPN = virt_to_phys(dsc->ring_state) >> PAGE_SHIFT; > + cmd.reqRingPPNs[0] = virt_to_phys(dsc->ring_reqs) >> PAGE_SHIFT;; > + cmd.cmpRingPPNs[0] = virt_to_phys(dsc->ring_cmps) >> PAGE_SHIFT;; 1. Just one semicolon is needed at the end. 2. I am no C professional, but why are the `memset()` calls above needed when the content does not seems to be used, when the pointer is casted to `unsigned int`? > + > + pvscsi_write_cmd_desc(iobase, PVSCSI_CMD_SETUP_RINGS, > + &cmd, sizeof(cmd)); > + *ring_dsc = dsc; > +} […] Thanks, Paul

2 1

Re: [SeaBIOS] int09h_handler(): unknown scancode read: 0x5c!
by Kevin O'Connor 14 Oct '13

14 Oct '13

On Mon, Oct 14, 2013 at 12:35:53PM +0000, Stojsavljevic, Zoran wrote: > I am resending this email since it never reached the wide audience. > > Hello to seabios list (and to direct participants of below thread), > I have problem to boot to GRUB from SeaBIOS, since I was able to > overcome some of the obstacles on my way to here. Now I am blocked > with the KBD: int09h_handler(): unknown scancode read: 0x5c! Your log indicates you are using coreboot and have compiled seabios for qemu. You need to compile seabios for coreboot. -Kevin

2 2

[PATCH] qemu: handle e820 ram entries
by Gerd Hoffmann 14 Oct '13

14 Oct '13

Check the e820 entries we get from qemu for ram entries and handle them accordingly: (1) Entries overlapping with the lowmem range will be ignored. At least initially qemu will not send such entries for compatibility reasons. That may change in the future though if qemu needs a more recent seabios versions anyway for some reason. So better be prepared. (2) Adjust RamSizeOver4G variable if we find it being too small. Needed for guests with more that 1TB (this is the max amount of memory the three-byte cmmos interface can represent). Signed-off-by: Gerd Hoffmann <kraxel(a)redhat.com> --- src/fw/paravirt.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/src/fw/paravirt.c b/src/fw/paravirt.c index c118480..c1469ae 100644 --- a/src/fw/paravirt.c +++ b/src/fw/paravirt.c @@ -285,6 +285,15 @@ qemu_cfg_legacy(void) int i; for (i = 0; i < count32; i++) { qemu_cfg_read(&entry, sizeof(entry)); + if (entry.type == E820_RAM) { + if (entry.address < RamSize) + // ignore, we got it from cmos already and + // adding this again would ruin any reservations + // done so far + continue; + if (0x100000000LL + RamSizeOver4G < entry.address + entry.length) + RamSizeOver4G = entry.address + entry.length - 0x100000000LL; + } add_e820(entry.address, entry.length, entry.type); } } else if (runningOnKVM()) { -- 1.8.3.1

2 6

int09h_handler(): unknown scancode read: 0x5c!
by Stojsavljevic, Zoran 14 Oct '13

14 Oct '13

I am resending this email since it never reached the wide audience. Hello to seabios list (and to direct participants of below thread), I have problem to boot to GRUB from SeaBIOS, since I was able to overcome some of the obstacles on my way to here. Now I am blocked with the KBD: int09h_handler(): unknown scancode read: 0x5c! I was researching/goggling around, and I was able to determine the main cause why KBD does not work. Here is the update. I found the thread: http://www.seabios.org/pipermail/seabios/2012-March/003288.html Unfortunately, the SeaBIOS code has been changed. In src/usb-ohci.c I could not find any more ohci_free_pipe(), there is only ohci_free_pipes(), so I am not able to apply patch: diff --git a/src/usb-ohci.c b/src/usb-ohci.c index 7a437ad..4e020cb 100644 The reason (my best guess) is that from march 2012 to Present SeaBIOS changed significantly. Another patch which converts alloc_low() to alloc_high() (3 calls) I found and applied, but so far it did not work. Could anybody advice what I should do to overcome this problem? Please, find complete PyTTY.log file attached. Thank you, Zoran _______ Most of The Time you should be "intel inside" to be capable to think "out of the box". Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen, Deutschland Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas Lusk Registergericht: Muenchen HRB 47456 Ust.-IdNr./VAT Registration No.: DE129385895 Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052

1 0

Re: [SeaBIOS] [PATCH v2] pvscsi boot support
by Kevin O'Connor 13 Oct '13

13 Oct '13

On Thu, Oct 10, 2013 at 10:48:19AM +0300, Evgeny Budilovsky wrote: > On Thu, Oct 10, 2013 at 3:30 AM, Kevin O'Connor <kevin(a)koconnor.net> wrote: > > > On Mon, Oct 07, 2013 at 01:16:30PM +0300, Evgeny Budilovsky wrote: > > > > > > Signed-off-by: Evgeny Budilovsky <evgeny.budilovsky(a)ravellosystems.com> > > > > Thanks for submitting. > > > > I'm fine with the patch. But can you expand on what "pvscsi" is, what > > advantages there is for an user to enable it, what hypervisors support > > it, etc. Ideally this info would be in the Kconfig menu as well. > > > Hi, > I was thinking about writing similar to this in the Kconfig help item. What > do you think ? > If this is ok I will resend the patch Yes - looks good to me. -Kevin

2 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS October 2013