SeaBIOS

seabios@seabios.org

3770 discussions

[PATCH v3 0/3] esp-scsi: fix reliance on QEMU ESP SCSI implementation bugs
by Mark Cave-Ayland 24 Aug '23

24 Aug '23

Whilst trying to improve QEMU's ESP SCSI implementation with regard to the NCR datasheet, I noticed that SeaBIOS would fail to boot one of my Debian test images with my patches applied. Further investigation revealed that the SeaBIOS esp-scsi driver inadvertently relies on existing bugs in QEMU's ESP SCSI emulation to work correctly, so this series changes the driver to work as described in the NCR datasheet. With this series applied it is possible for the updated SeaBIOS to boot both QEMU current git HEAD as well as my local WIP branch containing various fixes and improvements to QEMU's ESP SCSI emulation. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> v3: - Add R-B tags for patches 1 and 2 from Paolo - Add patch 3 to handle another bug to be fixed in QEMU's ESP emulation v2: - Fix typo in patch 2 commit message - Add reference to datasheet in patch 2 commit message as requested by Paul Mark Cave-Ayland (3): esp-scsi: flush FIFO before sending SCSI command esp-scsi: check for INTR_BS/INTR_FC instead of STAT_TC for command completion esp-scsi: handle non-DMA SCSI commands with no data phase src/hw/esp-scsi.c | 49 +++++++++++++++++++++++++++++++++-------------- 1 file changed, 35 insertions(+), 14 deletions(-) -- 2.39.2

3 5

[Increase BUILD_MAX_E820 to 128] Increase BUILD_MAX_E820 to 128
by Tony Titus 24 Aug '23

24 Aug '23

For platforms with high number of numa nodes, 32 e820 entries are not enough. Linux kernel sets the maximum e820 entries to a base value of 128. Setting BUILD_MAX_E820 to 128 to be in sync with this base value. Signed-off-by: Tony Titus <tonydt(a)amazon.com> --- src/config.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/config.h b/src/config.h index 9abe355..16952eb 100644 --- a/src/config.h +++ b/src/config.h @@ -15,7 +15,7 @@ #define BUILD_APPNAME4 "BXPC" // Maximum number of map entries in the e820 map -#define BUILD_MAX_E820 32 +#define BUILD_MAX_E820 128 // Space to reserve in high-memory for tables #define BUILD_MIN_HIGHTABLE (256*1024) #define BUILD_MAX_HIGHTABLE (16*1024*1024) -- 2.40.1

2 1

[PATCH v2 0/2] esp-scsi: fix reliance on QEMU ESP SCSI implementation bugs
by Mark Cave-Ayland 07 Aug '23

07 Aug '23

Whilst trying to improve QEMU's ESP SCSI implementation with regard to the NCR datasheet, I noticed that SeaBIOS would fail to boot one of my Debian test images with my patches applied. Further investigation revealed that the SeaBIOS esp-scsi driver inadvertently relies on existing bugs in QEMU's ESP SCSI emulation to work correctly, so this series changes the driver to work as described in the NCR datasheet. With this series applied it is possible for the updated SeaBIOS to boot both QEMU current git HEAD as well as my local WIP branch containing various fixes and improvements to QEMU's ESP SCSI emulation. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> v2: - Fix typo in patch 2 commit message - Add reference to datasheet in patch 2 commit message as requested by Paul Mark Cave-Ayland (2): esp-scsi: flush FIFO before sending SCSI command esp-scsi: check for INTR_BS/INTR_FC instead of STAT_TC for command completion src/hw/esp-scsi.c | 38 +++++++++++++++++++++++++------------- 1 file changed, 25 insertions(+), 13 deletions(-) -- 2.39.2

3 5

[PATCH 0/2] esp-scsi: fix reliance on QEMU ESP SCSI implementation bugs
by Mark Cave-Ayland 02 Aug '23

02 Aug '23

Whilst trying to improve QEMU's ESP SCSI implementation with regard to the NCR datasheet, I noticed that SeaBIOS would fail to boot one of my Debian test images with my patches applied. Further investigation revealed that the SeaBIOS esp-scsi driver inadvertently relies on existing bugs in QEMU's ESP SCSI emulation to work correctly, so this series changes the driver to work as described in the NCR datasheet. With this series applied it is possible for the updated SeaBIOS to boot both QEMU current git HEAD as well as my local WIP branch containing various fixes and improvements to QEMU's ESP SCSI emulation. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Mark Cave-Ayland (2): esp-scsi: flush FIFO before sending SCSI command esp-scsi: check for INTR_BS/INTR_FC instead of STAT_TC for command completion src/hw/esp-scsi.c | 38 +++++++++++++++++++++++++------------- 1 file changed, 25 insertions(+), 13 deletions(-) -- 2.30.2

3 7

[PATCH] paravirt: call pci_bios_init_devices() when runningOnXen()
by Xenia Ragiadakou 14 Jul '23

14 Jul '23

The issue that this patch attempts to address is S3 related. Currently, suspending a XEN guest multiple times does not work. This happens because PIIX4 PM io space gets unmapped during S3 resume and any accesses performed to trigger subsequent suspends are not handled. So, the guest spins on the wake sts bit forever. Below follows a more detailed description of the issue. On S3 resume path, QEMU issues piix4_pm_reset() to bring PPIX4 PM controller to the state expected. More specifically, piix4_pm_reset() zeroes PMIOSE register, i.e pci_conf[0x80] = 0, and this results in unmapping PIIX4 PM io space, i.e. memory_region_set_enabled(&s->io, d->config[0x80] & 1) It is bios responsibility to reenable PIIX4 PM io space during S3 resume path. However, when running on XEN, the qemu platform initialization code does not call pci_bios_init_devices(), hence the PIIX4 PM controller init function is not called and PiixPmBDF is not set. The value of PiixPmBDF indicates the presence of PIIX4 PM controller. Thus, S3 resume code in pci_resume() considers that PIIX4 PM controller is not present and does not perform the relevant post-resume setup that restores PIIX4 PM io space. This is the reason why any subsequent guest access to those registers is not handled. This patch adds a call to pci_bios_init_devices() to the early xen platform setup code. Signed-off-by: Xenia Ragiadakou <xenia.ragiadakou(a)amd.com> --- src/fw/paravirt.c | 1 + src/fw/pciinit.c | 2 +- src/util.h | 1 + 3 files changed, 3 insertions(+), 1 deletion(-) diff --git a/src/fw/paravirt.c b/src/fw/paravirt.c index fba4e52..0c035eb 100644 --- a/src/fw/paravirt.c +++ b/src/fw/paravirt.c @@ -205,6 +205,7 @@ qemu_platform_setup(void) if (runningOnXen()) { pci_probe_devices(); + pci_bios_init_devices(); xen_hypercall_setup(); xen_biostable_setup(); return; diff --git a/src/fw/pciinit.c b/src/fw/pciinit.c index badf13d..f955481 100644 --- a/src/fw/pciinit.c +++ b/src/fw/pciinit.c @@ -417,7 +417,7 @@ static void pci_bios_init_device(struct pci_device *pci) PCI_BRIDGE_CTL_SERR); } -static void pci_bios_init_devices(void) +void pci_bios_init_devices(void) { struct pci_device *pci; foreachpci(pci) { diff --git a/src/util.h b/src/util.h index aff8e88..ebd863f 100644 --- a/src/util.h +++ b/src/util.h @@ -137,6 +137,7 @@ extern u64 pcimem64_start, pcimem64_end; extern const u8 pci_irqs[4]; void pci_setup(void); void pci_resume(void); +void pci_bios_init_devices(void); // fw/pirtable.c void pirtable_setup(void); -- 2.34.1

1 0

[PATCH] ahci: handle TFES irq correctly
by Niklas Cassel 07 Jul '23

07 Jul '23

According to AHCI 1.3.1, 5.3.8.1 RegFIS:Entry, if ERR_STAT is set in the received FIS, the HBA shall jump to state ERR:FatalTaskfile, which will raise a TFES IRQ. This means that if ERR_STAT is set in the recevied FIS, PxIS.TFES will be set, without either PxIS.DHRS or PxIS.PSS being set. SeaBIOS function ahci_port_setup() will try to identify an AHCI device by sending an ATAPI identify device command. However, such a command will be aborted with ERR_STAT set for a regular (non-ATAPI) device. ahci_command() already performs the correct error recovery steps when status is correctly set, so simply modify ahci_command() to read the correct status when PxIS.TFES is set. It is safe to read PxTFD when PxIS.TFES is set, even for systems with a port multiplier, see AHCI 1.3.1, 9.3.7 PxTFD Register Information: "When a taskfile error occurs (PxIS.TFES is set to '1'), the host may refer to the values in PxTFD. The values in PxTFD at this time are guaranteed to correspond to the device that reported the taskfile error condition." Without this, each boot will be delayed by 32 seconds, waiting for the AHCI command to timeout. Signed-off-by: Niklas Cassel <niklas.cassel(a)wdc.com> --- src/hw/ahci.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/hw/ahci.c b/src/hw/ahci.c index d45b430..3fa845a 100644 --- a/src/hw/ahci.c +++ b/src/hw/ahci.c @@ -138,6 +138,12 @@ static int ahci_command(struct ahci_port_s *port_gf, int iswrite, int isatapi, intbits = ahci_port_readl(ctrl, pnr, PORT_IRQ_STAT); if (intbits) { ahci_port_writel(ctrl, pnr, PORT_IRQ_STAT, intbits); + if (intbits & 0x40000000) { + u32 tf = ahci_port_readl(ctrl, pnr, PORT_TFDATA); + status = tf & 0xff; + error = (tf & 0xff00) >> 8; + break; + } if (intbits & 0x02) { status = GET_LOWFLAT(fis->psfis[2]); error = GET_LOWFLAT(fis->psfis[3]); -- 2.40.1

5 5

[PATCH] git: Add .mailmap entries for patches claiming list authorship
by Philippe Mathieu-Daudé 07 Jul '23

07 Jul '23

Similarly to QEMU's commit 3bd2608db7 ("maint: Add .mailmap entries for patches claiming list authorship") [*], add a .mailmap file to SeaBIOS repository to fix the following commits: - 1281e34 ("ahci: handle TFES irq correctly") authored by Niklas Cassel <niklas.cassel(a)wdc.com> - cd93345 ("virtio-blk: Fix integer overflow for large max IO sizes") authored by Lukas Stockner <lstockner(a)genesiscloud.com> - dc776a2 ("nvme: avoid use-after-free in nvme_controller_enable()") authored by Jan Beulich <jbeulich(a)suse.com> - 54082c8 ("nvme: fix missing newline on sq full print") authored by Alex Martens <alexmgit(a)protonmail.com> - b0d61ec ("usb-hid: Increase MAX_KBD_EVENT") authored by Stefan Ott <stefan(a)ott.net> [*] https://gitlab.com/qemu-project/qemu/-/commit/3bd2608db72 Fixes: 1281e34 ("ahci: handle TFES irq correctly") Fixes: cd93345 ("virtio-blk: Fix integer overflow for large max IO sizes") Fixes: dc776a2 ("nvme: avoid use-after-free in nvme_controller_enable()") Fixes: 54082c8 ("nvme: fix missing newline on sq full print") Fixes: b0d61ec ("usb-hid: Increase MAX_KBD_EVENT") Reported-by: Paul Menzel <pmenzel(a)molgen.mpg.de> Signed-off-by: Philippe Mathieu-Daudé <philmd(a)linaro.org> --- .mailmap | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) create mode 100644 .mailmap diff --git a/.mailmap b/.mailmap new file mode 100644 index 0000000..1411ae5 --- /dev/null +++ b/.mailmap @@ -0,0 +1,23 @@ +# This mailmap fixes up author names/addresses. +# +# If you are adding to this file consider if a similar change needs to +# be made to contrib/gitdm/aliases. They are not however completely +# analogous. .mailmap is concerned with fixing up damaged author +# fields where as the gitdm equivalent is more concerned with making +# sure multiple email addresses get mapped onto the same author. +# +# From man git-shortlog the forms are: +# +# Proper Name <commit(a)email.xx> +# <proper(a)email.xx> <commit(a)email.xx> +# Proper Name <proper(a)email.xx> <commit(a)email.xx> +# Proper Name <proper(a)email.xx> Commit Name <commit(a)email.xx> +# + +# Translate a few commits where mailman rewrote the From: line due to +# strict SPF, although we prefer to avoid adding more entries like that. +Niklas Cassel <niklas.cassel(a)wdc.com> Niklas Cassel via SeaBIOS <seabios(a)seabios.org> +Lukas Stockner <lstockner(a)genesiscloud.com> Lukas Stockner via SeaBIOS <seabios(a)seabios.org> +Jan Beulich <jbeulich(a)suse.com> Jan Beulich via SeaBIOS <seabios(a)seabios.org> +Alex Martens <alexmgit(a)protonmail.com> Alex Martens via SeaBIOS <seabios(a)seabios.org> +Stefan Ott <stefan(a)ott.net> Stefan Ott via SeaBIOS <seabios(a)seabios.org> -- 2.38.1

1 0

PATCH: search PCI option ROMs on CBFS in case insensitive way
by Leah Rowe 13 Jun '23

13 Jun '23

Hi, Attached is a patch for SeaBIOS's searching of CBFS, specifically for PCI option ROMs as per PCI vendor/device ID detected in each PCI device scanned. At present, SeaBIOS assumes that the hexademical characters in vendor/device ID are all-lowercase, which is not necessarily a problem, but it may not be obvious to users. Sure enough, I had a user earlier who couldn't get their Nvidia card working properly even though they had the correct VGA ROM. After much trial and error, the user realised that they needed the PCI ROM to have an all-lowercase file name in CBFS. I suspect that a lot of users will make the same mistake, and not come to the same realisation as the other user. To prevent such headaches, the attached patch modifies SeaBIOS's handling such that PCI option ROM file names are treated as case insensitive, when scanning CBFS for files in the notation: pcivvvv,dddd.rom -- Leah Rowe <leah(a)libreboot.org>

2 1

PATCH: properly cast c value in memset
by Leah Rowe 13 Jun '23

13 Jun '23

Hi Attached is a patch I did earlier, while working on something else. I noticed that SeaBIOS's memset() implementation casts the object "s" to (char *), which is technically incorrect; see: https://pubs.opengroup.org/onlinepubs/9699919799/functions/memset.html It says: The memset() function shall copy c (converted to an unsigned char) into each of the first n bytes of the object pointed to by s. The attached patch fixes this, by casting c to unsigned char, and accordingly, casts void *s to (unsigned char *) The current cast does (char *) which is technically incorrect, because whether char is signed or unsigned is implementation-defined, so you have to explicitly state signed or unsigned when signedness matters. -- Leah Rowe <leah(a)libreboot.org>

2 1

[PATCH] virtio-blk: Fix integer overflow for large max IO sizes
by Lukas Stockner 13 Jun '23

13 Jun '23

When the maximum IO size supported by the virtio-blk backend is large enough (>= 32MiB for 512B sectors), the computed blk_num_max will overflow. In particular, if it's a multiple of 32MiB, blk_num_max will end up as zero, causing IO requests to fail. This is triggered by e.g. the SPDK virtio-blk vhost-user backend. To fix it, just limit blk_num_max to 65535 before converting to u16. Signed-off-by: Lukas Stockner <lstockner(a)genesiscloud.com> --- src/hw/virtio-blk.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/hw/virtio-blk.c b/src/hw/virtio-blk.c index e087fe4..137a2c3 100644 --- a/src/hw/virtio-blk.c +++ b/src/hw/virtio-blk.c @@ -92,7 +92,7 @@ virtio_blk_op(struct disk_op_s *op, int write) u16 blk_num_max; if (vdrive->drive.blksize != 0 && max_io_size != 0) - blk_num_max = (u16)(max_io_size / vdrive->drive.blksize); + blk_num_max = (u16) min(max_io_size / vdrive->drive.blksize, 0xffff); else /* default blk_num_max if hardware doesnot advise a proper value */ blk_num_max = 64; -- 2.41.0

2 1

← Newer
1
2
3
4
5
6
7
8
...
377
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS