SeaBIOS

seabios@seabios.org

4 participants
3800 discussions

Start a nNew thread

seabios fails to build with current IASL
by Michael Tokarev 15 Jul '18

15 Jul '18

Here's the debian bugreport https://bugs.debian.org/901462 with the patch. Thanks! /mjt

3 5

seabios fails to build with current IASL
by Vivia Nikolaidou 12 Jul '18

12 Jul '18

Hi, Thanks for taking action on this bug that I reported. This is a follow-up from https://mail.coreboot.org/pipermail/seabios/2018-July/012353.html - I just subscribed to the mailing list and I don't have an easy way to reply to that email directly. On Mon, Jul 02, 2018 at 09:10:34AM +0200, Gerd Hoffmann wrote: > > If you don't want use the prebuilt tables I think the sane options are > (a) use a older iasl compiler, or (b) flip ACPI_DSDT to "n" and drop > support for qemu 1.3 + older. Since (a) isn't an option in Debian, I tried going for (b) (did it directly in debian/rules when generating .config, without adding a patch for it). It still failed to compile without the patch that truncates the Table ID in the DefinitionBlock. Removing the prebuilt tables before compilation also didn't work, as they are used quite extensively in acpi.c. I'm not sure I haven't misunderstood something here, but when you say that flipping ACPI_DSDT to "n" would make it not use the prebuilt tables, what exactly do you mean? That said, would the combination of the truncating patch + flipping ACPI_DSDT to "n" be an acceptable solution to fix the build in Debian? Best regards, Vivia

1 0

[PATCH] virtio-blk/scsi: enable multi-queues support when starting device
by Changpeng Liu 10 Jul '18

10 Jul '18

QEMU will not start all the queues since commit fb20fbb76 "vhost: avoid to start/stop virtqueue which is not read", because seabios only use one queue when starting, this will not work for some vhost slave targets which expect the exact number of queues defined in virtio-pci configuration space, while here, we also enable those queues in the BIOS phase. Signed-off-by: Changpeng Liu <changpeng.liu(a)intel.com> --- src/hw/virtio-blk.c | 26 +++++++++++++++++++------- src/hw/virtio-ring.h | 1 + src/hw/virtio-scsi.c | 28 +++++++++++++++++++--------- 3 files changed, 39 insertions(+), 16 deletions(-) diff --git a/src/hw/virtio-blk.c b/src/hw/virtio-blk.c index 88d7e54..79638ec 100644 --- a/src/hw/virtio-blk.c +++ b/src/hw/virtio-blk.c @@ -25,7 +25,7 @@ struct virtiodrive_s { struct drive_s drive; - struct vring_virtqueue *vq; + struct vring_virtqueue *vq[MAX_NUM_QUEUES]; struct vp_device vp; }; @@ -34,7 +34,7 @@ virtio_blk_op(struct disk_op_s *op, int write) { struct virtiodrive_s *vdrive = container_of(op->drive_fl, struct virtiodrive_s, drive); - struct vring_virtqueue *vq = vdrive->vq; + struct vring_virtqueue *vq = vdrive->vq[0]; struct virtio_blk_outhdr hdr = { .type = write ? VIRTIO_BLK_T_OUT : VIRTIO_BLK_T_IN, .ioprio = 0, @@ -96,6 +96,7 @@ virtio_blk_process_op(struct disk_op_s *op) static void init_virtio_blk(void *data) { + u32 i, num_queues = 1; struct pci_device *pci = data; u8 status = VIRTIO_CONFIG_S_ACKNOWLEDGE | VIRTIO_CONFIG_S_DRIVER; dprintf(1, "found virtio-blk at %pP\n", pci); @@ -109,10 +110,6 @@ init_virtio_blk(void *data) vdrive->drive.cntl_id = pci->bdf; vp_init_simple(&vdrive->vp, pci); - if (vp_find_vq(&vdrive->vp, 0, &vdrive->vq) < 0 ) { - dprintf(1, "fail to find vq for virtio-blk %pP\n", pci); - goto fail; - } if (vdrive->vp.use_modern) { struct vp_device *vp = &vdrive->vp; @@ -156,6 +153,11 @@ init_virtio_blk(void *data) vp_read(&vp->device, struct virtio_blk_config, heads); vdrive->drive.pchs.sector = vp_read(&vp->device, struct virtio_blk_config, sectors); + + num_queues = vp_read(&vp->common, virtio_pci_common_cfg, num_queues); + if (num_queues < 1 || num_queues > MAX_NUM_QUEUES) { + num_queues = 1; + } } else { struct virtio_blk_config cfg; vp_get_legacy(&vdrive->vp, 0, &cfg, sizeof(cfg)); @@ -178,6 +180,13 @@ init_virtio_blk(void *data) vdrive->drive.pchs.sector = cfg.sectors; } + for (i = 0; i < num_queues; i++) { + if (vp_find_vq(&vdrive->vp, i, &vdrive->vq[i]) < 0 ) { + dprintf(1, "fail to find vq %u for virtio-blk %pP\n", i, pci); + goto fail_vq; + } + } + char *desc = znprintf(MAXDESCSIZE, "Virtio disk PCI:%pP", pci); boot_add_hd(&vdrive->drive, desc, bootprio_find_pci_device(pci)); @@ -185,9 +194,12 @@ init_virtio_blk(void *data) vp_set_status(&vdrive->vp, status); return; +fail_vq: + for (i = 0; i < num_queues; i++) { + free(vdrive->vq[i]); + } fail: vp_reset(&vdrive->vp); - free(vdrive->vq); free(vdrive); } diff --git a/src/hw/virtio-ring.h b/src/hw/virtio-ring.h index 8604a01..3c8a2d1 100644 --- a/src/hw/virtio-ring.h +++ b/src/hw/virtio-ring.h @@ -21,6 +21,7 @@ #define VIRTIO_F_IOMMU_PLATFORM 33 #define MAX_QUEUE_NUM (128) +#define MAX_NUM_QUEUES (128) #define VRING_DESC_F_NEXT 1 #define VRING_DESC_F_WRITE 2 diff --git a/src/hw/virtio-scsi.c b/src/hw/virtio-scsi.c index a87cad8..d08643f 100644 --- a/src/hw/virtio-scsi.c +++ b/src/hw/virtio-scsi.c @@ -148,9 +148,10 @@ virtio_scsi_scan_target(struct pci_device *pci, struct vp_device *vp, static void init_virtio_scsi(void *data) { + u32 i, num_queues = 3; struct pci_device *pci = data; dprintf(1, "found virtio-scsi at %pP\n", pci); - struct vring_virtqueue *vq = NULL; + struct vring_virtqueue *vq[MAX_NUM_QUEUES]; struct vp_device *vp = malloc_high(sizeof(*vp)); if (!vp) { warn_noalloc(); @@ -175,29 +176,38 @@ init_virtio_scsi(void *data) dprintf(1, "device didn't accept features: %pP\n", pci); goto fail; } + + num_queues = vp_read(&vp->common, virtio_pci_common_cfg, num_queues); + if (num_queues < 3 || num_queues > MAX_NUM_QUEUES) { + num_queues = 3; + } } - if (vp_find_vq(vp, 2, &vq) < 0 ) { - dprintf(1, "fail to find vq for virtio-scsi %pP\n", pci); - goto fail; + for (i = 0; i < num_queues; i++) { + if (vp_find_vq(vp, i, &vq[i]) < 0 ) { + dprintf(1, "fail to find vq %u for virtio-scsi %pP\n", i, pci); + goto fail; + } } status |= VIRTIO_CONFIG_S_DRIVER_OK; vp_set_status(vp, status); - int i, tot; + int tot; for (tot = 0, i = 0; i < 256; i++) - tot += virtio_scsi_scan_target(pci, vp, vq, i); + tot += virtio_scsi_scan_target(pci, vp, vq[2], i); if (!tot) - goto fail; + goto fail_vq; return; - +fail_vq: + for (i = 0; i < num_queues; i++) { + free(vq[i]); + } fail: vp_reset(vp); free(vp); - free(vq); } void -- 1.9.3

1 0

[PATCH 1/2] nvme: add option to toggle probing only boot devices
by Julian Stecklina 06 Jul '18

06 Jul '18

Some systems have a large number of NVMe controllers. Probing all of them requires an excessive amount of memory and with out-of-order probing this also means that the important ones will not be probed, because SeaBIOS runs out of memory. Add an option to skip probing of NVMe devices, if there is no boot priority for them. Signed-off-by: Julian Stecklina <jsteckli(a)amazon.de> --- src/Kconfig | 8 ++++++++ src/hw/nvme.c | 14 +++++++++++++- 2 files changed, 21 insertions(+), 1 deletion(-) diff --git a/src/Kconfig b/src/Kconfig index 55a87cb..ef8f458 100644 --- a/src/Kconfig +++ b/src/Kconfig @@ -233,6 +233,14 @@ menu "Hardware support" default y help Support for NVMe disk code. + config NVME_PROBE_BOOTDEVICES_ONLY + depends on NVME + bool "Only probe NVMe controllers with boot priority" + default n + help + Only enables NVMe controllers that have a boot priority + configured. This saves memory on systems with many non-bootable + NVMe devices. config PS2PORT depends on KEYBOARD || MOUSE diff --git a/src/hw/nvme.c b/src/hw/nvme.c index e6d739d..5525f68 100644 --- a/src/hw/nvme.c +++ b/src/hw/nvme.c @@ -625,6 +625,16 @@ nvme_controller_setup(void *opaque) dprintf(2, "Failed to enable NVMe controller.\n"); } + +static int should_setup_controller(struct pci_device *pci) +{ +#ifdef NVME_PROBE_BOOTDEVICES_ONLY + return bootprio_find_pci_device(pci) >= 0; +#else + return 1; +#endif +} + // Locate and init NVMe controllers static void nvme_scan(void) @@ -640,7 +650,9 @@ nvme_scan(void) continue; } - run_thread(nvme_controller_setup, pci); + if (should_setup_controller(pci)) { + run_thread(nvme_controller_setup, pci); + } } } -- 2.7.4

4 7

[PATCH 0/7] seabios: 1.11.2 update
by Gerd Hoffmann 02 Jul '18

02 Jul '18

Hi folks, So, here is the current state of the seabios update for the upcoming qemu 3.0 release. This time it is a bit more than just the usual "update submodule and binaries". seabios 1.11.2 is not released yet. The plan is to cherry-pick the changes needed to support the new qemu 3.0 display devices (bochs-display and ramfb) into the 1.11-stable branch for the 1.11.2 release. Current state is here: https://git.kraxel.org/cgit/seabios/log/?h=1.11-stable-testing Changes in this patch series: - seabios submodule update (to the testing branch). - new vgabios configs for bochs-display and ramfb. - update the bios config to enables ide dma (for qnx). - drop (old) vgabios submodule and build rules. - seabios binary update. - make bochs-display and ramfb load the new vgabios roms. Git branch for this series is here: https://git.kraxel.org/cgit/qemu/log/?h=seabios-1.11.2 cheers, Gerd Gerd Hoffmann (7): seabios: update submodule to 1.11.2-testing seabios: add vga configs for bochs-display and ramfb seabios: enable ide dma vgabios: remove submodule and build rules. seabios: update binaries bochs-display: enable vgabios ramfb: enable vgabios hw/display/bochs-display.c | 1 + hw/display/ramfb.c | 1 + .gitmodules | 3 --- pc-bios/bios-256k.bin | Bin 262144 -> 262144 bytes pc-bios/bios.bin | Bin 131072 -> 131072 bytes pc-bios/vgabios-cirrus.bin | Bin 38400 -> 38400 bytes pc-bios/vgabios-qxl.bin | Bin 38912 -> 38912 bytes pc-bios/vgabios-stdvga.bin | Bin 38912 -> 38912 bytes pc-bios/vgabios-virtio.bin | Bin 38912 -> 38912 bytes pc-bios/vgabios-vmware.bin | Bin 38912 -> 38912 bytes pc-bios/vgabios.bin | Bin 38400 -> 38400 bytes roms/Makefile | 20 +++----------------- roms/config.seabios-128k | 1 + roms/config.seabios-256k | 1 + roms/config.vga-bochs-display | 3 +++ roms/config.vga-ramfb | 3 +++ roms/seabios | 2 +- roms/vgabios | 1 - 18 files changed, 14 insertions(+), 22 deletions(-) create mode 100644 roms/config.vga-bochs-display create mode 100644 roms/config.vga-ramfb delete mode 160000 roms/vgabios -- 2.9.3

1 7

seabios issue
by Roger Lawhorn 01 Jul '18

01 Jul '18

Hello, Recently my qemu with gpu passthrough setup quit working. I think it is seabios related. i have turned on debugging for seabios. Here is the output from seabios from a successful windows 10 boot using an emulated vga adapter: https://pastebin.com/f3DCqTbc Here is the output when using my nvidia gtx 880m 8gb gaming card: https://pastebin.com/GpWKZadh when seabios runs the option rom for nvidia it just halts. i have never debuged seabios before so i dont know what would normally show up at this point. i.e. is this normal or is something wrong. during the vga boot all 8 cpus go to 100% for a short time then drop. during the failed boot with nvidia 1 cpu goes to 100% and stays there forever. i have let it run all night at times. i am seeking to figure out if i have a seabios issue. i am running linux mint 18.3 and the repo copy of seabios is 1.8.2. i have booted this 100's of times and never had an issue before. thanks

2 2

ACPI OEM Table ID length
by coypu＠sdf.org 28 Jun '18

28 Jun '18

Hi folks, I ran into this problem while building Xen. unfortunately, I have trouble building the same files with seabios git, and have even more limited ability to test the changes. Apparently ACPI says that the OEM Table Id must be length 8. Some of the tables in seabios use longer. newer iASL rejects this and fails to build the code. Most likely things handle it as if it was truncated, but perhaps it overran to the next field. I've used this diff, for your consideration. diff --git a/src/fw/ssdt-misc.dsl b/src/fw/ssdt-misc.dsl index acc850e..d1d2c9e 100644 --- a/src/fw/ssdt-misc.dsl +++ b/src/fw/ssdt-misc.dsl @@ -1,6 +1,6 @@ ACPI_EXTRACT_ALL_CODE ssdp_misc_aml -DefinitionBlock ("ssdt-misc.aml", "SSDT", 0x01, "BXPC", "BXSSDTSUSP", 0x1) +DefinitionBlock ("ssdt-misc.aml", "SSDT", 0x01, "BXPC", "BXSSDTSU", 0x1) { /**************************************************************** diff --git a/src/fw/ssdt-pcihp.dsl b/src/fw/ssdt-pcihp.dsl index cb24c11..518a5eb 100644 --- a/src/fw/ssdt-pcihp.dsl +++ b/src/fw/ssdt-pcihp.dsl @@ -1,6 +1,6 @@ ACPI_EXTRACT_ALL_CODE ssdp_pcihp_aml -DefinitionBlock ("ssdt-pcihp.aml", "SSDT", 0x01, "BXPC", "BXSSDTPCIHP", 0x1) +DefinitionBlock ("ssdt-pcihp.aml", "SSDT", 0x01, "BXPC", "BXSSDTPC", 0x1) { /****************************************************************

1 0

[PATCH] cbvga_set_mode: refine clear display logic
by Gerd Hoffmann 27 Jun '18

27 Jun '18

There is no need to skip text mode clearing. Clearing the framebuffer should be safe on legacy calls too. Signed-off-by: Gerd Hoffmann <kraxel(a)redhat.com> --- vgasrc/cbvga.c | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-) diff --git a/vgasrc/cbvga.c b/vgasrc/cbvga.c index f6ebe71242..859524cb23 100644 --- a/vgasrc/cbvga.c +++ b/vgasrc/cbvga.c @@ -201,18 +201,20 @@ cbvga_set_mode(struct vgamode_s *vmode_g, int flags) */ u8 extra_stack = GET_BDA_EXT(flags) & BF_EXTRA_STACK; MASK_BDA_EXT(flags, BF_EMULATE_TEXT, emul ? BF_EMULATE_TEXT : 0); - if (!(flags & MF_NOCLEARMEM) && extra_stack) { + if (!(flags & MF_NOCLEARMEM)) { if (GET_GLOBAL(CBmodeinfo.memmodel) == MM_TEXT) { memset16_far(SEG_CTEXT, (void*)0, 0x0720, 80*25*2); return 0; } - struct gfx_op op; - init_gfx_op(&op, &CBmodeinfo); - op.x = op.y = 0; - op.xlen = GET_GLOBAL(CBmodeinfo.width); - op.ylen = GET_GLOBAL(CBmodeinfo.height); - op.op = GO_MEMSET; - handle_gfx_op(&op); + if (extra_stack || flags & MF_LEGACY) { + struct gfx_op op; + init_gfx_op(&op, &CBmodeinfo); + op.x = op.y = 0; + op.xlen = GET_GLOBAL(CBmodeinfo.width); + op.ylen = GET_GLOBAL(CBmodeinfo.height); + op.op = GO_MEMSET; + handle_gfx_op(&op); + } } return 0; } -- 2.9.3

3 4

Re: [SeaBIOS] [PATCH v2 6/6] cbvga_set_mode: disable clearmem in windows x86 emulator.
by Patrick Rudolph 26 Jun '18

26 Jun '18

On Mon, 2018-06-04 at 09:31 -0400, Kevin O'Connor wrote: > On Mon, Jun 04, 2018 at 09:29:17AM +0200, Gerd Hoffmann wrote: > > Signed-off-by: Gerd Hoffmann <kraxel(a)redhat.com> > > --- > > vgasrc/cbvga.c | 10 +++++++++- > > 1 file changed, 9 insertions(+), 1 deletion(-) > > > > diff --git a/vgasrc/cbvga.c b/vgasrc/cbvga.c > > index 3f16bee10c..f6ebe71242 100644 > > --- a/vgasrc/cbvga.c > > +++ b/vgasrc/cbvga.c > > @@ -192,8 +192,16 @@ int > > cbvga_set_mode(struct vgamode_s *vmode_g, int flags) > > { > > u8 emul = vmode_g == &CBemulinfo || GET_GLOBAL(CBmode) == 0x03; > > + /* > > + * The extra_stack flag is false when running in windows x86 > > + * emulator, to avoid stack switching triggering bugs. Using the > > + * same flag here to skip screen clearing, because the windows > > + * emulator seems to have problems to handle the int 1587 call > > + * too, and GO_MEMSET uses that. > > + */ > > + u8 extra_stack = GET_BDA_EXT(flags) & BF_EXTRA_STACK; > > MASK_BDA_EXT(flags, BF_EMULATE_TEXT, emul ? BF_EMULATE_TEXT : 0); > > - if (!(flags & MF_NOCLEARMEM)) { > > + if (!(flags & MF_NOCLEARMEM) && extra_stack) { > > FYI, extra_stack will only be true if CONFIG_VGA_ALLOCATE_EXTRA_STACK > is set, but I guess that's okay. > > Thanks, the series looks good to me. > -Kevin This commit breaks SeaBIOS' menu when used with coreboot's native graphic init in text mode. Payloads are not affacted (tested NTLDR and coreinfo). Everything works fine with native graphic init in high resolution framebuffer mode. I've got no debug log at hand, but I'll investigate asap. Tested on Lenovo T500 using libgfxinit with latest coreboot master. Regards, Patrick > > _______________________________________________ > SeaBIOS mailing list > SeaBIOS(a)seabios.org > https://mail.coreboot.org/mailman/listinfo/seabios

3 3

Re: [SeaBIOS] [PATCH v2 6/6] cbvga_set_mode: disable clearmem in windows x86 emulator.
by Patrick Rudolph 22 Jun '18

22 Jun '18

On Fri, 2018-06-22 at 10:14 -0400, Kevin O'Connor wrote: > On Fri, Jun 22, 2018 at 08:42:50AM +0200, Gerd Hoffmann wrote: > > On Wed, Jun 20, 2018 at 08:00:28PM +0200, Patrick Rudolph wrote: > > > This commit breaks SeaBIOS' menu when used with coreboot's native > > > graphic init in text mode. Payloads are not affacted (tested NTLDR and > > > coreinfo). > > > Everything works fine with native graphic init in high resolution > > > framebuffer mode. I've got no debug log at hand, but I'll investigate > > > asap. > > > Tested on Lenovo T500 using libgfxinit with latest coreboot master. > > > > Hmm, no clue what could cause this. But clearing the memory is only > > problematic for the framebuffer case, so maybe the patch below helps? > > > > diff --git a/vgasrc/cbvga.c b/vgasrc/cbvga.c > > index f6ebe71242..a48f4dc885 100644 > > --- a/vgasrc/cbvga.c > > +++ b/vgasrc/cbvga.c > > @@ -201,18 +201,20 @@ cbvga_set_mode(struct vgamode_s *vmode_g, int flags) > > */ > > u8 extra_stack = GET_BDA_EXT(flags) & BF_EXTRA_STACK; > > MASK_BDA_EXT(flags, BF_EMULATE_TEXT, emul ? BF_EMULATE_TEXT : 0); > > - if (!(flags & MF_NOCLEARMEM) && extra_stack) { > > + if (!(flags & MF_NOCLEARMEM)) { > > if (GET_GLOBAL(CBmodeinfo.memmodel) == MM_TEXT) { > > memset16_far(SEG_CTEXT, (void*)0, 0x0720, 80*25*2); > > return 0; > > } > > - struct gfx_op op; > > - init_gfx_op(&op, &CBmodeinfo); > > - op.x = op.y = 0; > > - op.xlen = GET_GLOBAL(CBmodeinfo.width); > > - op.ylen = GET_GLOBAL(CBmodeinfo.height); > > - op.op = GO_MEMSET; > > - handle_gfx_op(&op); > > + if (extra_stack) { > > I'm also not sure on the cause, but I think it should be safe to > change this to: > > if (extra_stack || flags & MF_LEGACY) { > Seems to be the correct solution, as BF_EXTRA_STACK is set after cbvga_set_mode and thus extra_stack is 0 on first cbvga_set_mode call. > because if this is a legacy call then it should be safe to write to > highmem even if the last call was a vesa modeset. > > -Kevin

1 0

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS