SeaBIOS December 2018

seabios@seabios.org

9 participants
15 discussions

Re: [SeaBIOS] [PATCH] ramdisk: search for all available floppy images instead of one
by Kevin O'Connor 06 Dec '18

06 Dec '18

On Sat, Nov 24, 2018 at 11:46:53PM +0300, Mike Banon wrote: > Dear Kevin, > > Please could you tell me, why the memory allocation functions like > malloc_tmphigh() are failing at do_boot() stage? (boot.c) . Because of > this "VARVERIFY32INIT" problem described in my letter above - > https://mail.coreboot.org/pipermail/seabios/2018-November/012575.html > - I rewrote all my code to allocate the RAM for one floppy only (space > equal to max floppy size, which could be used by any selected floppy) > . But when I'm trying to boot that selected floppy - it is failing, > because these floppies are LZMA compressed inside CBFS and to > decompress them another malloc_tmphigh() is needed (from > cbfs_copyfile() / coreboot.c) / Although this code compiles fine (no > VARVERIFY32INIT problem) - this malloc_tmphigh() function is always > failing and SeaBIOS is freezing. > > Spent all my weekend debugging this problem and really stuck, any help > will be highly appreciated Hi, Only the "post" stage is able to allocate ram. Once the "boot" phase starts executing, areas of memory are locked down, and it's possible for 3rd party software to become resident in various areas of ram. SeaBIOS thus can't touch those areas of ram or modify the e820 map. The VARVERIFY32INIT flag is there to catch this type of common error. There's some info on this in the docs at: https://www.seabios.org/Memory_Model#Memory_available_during_initialization and: https://www.seabios.org/Execution_and_code_flow So, any type of allocation would have to occur before the boot stage. It should be possible (though not necessarily easy) to perform the allocation in the map_floppy_drive() phase of the code. -Kevin

2 1

Re: [SeaBIOS] [PATCH v2] qemu: fast boot when linuxboot optionrom is used
by Kevin O'Connor 05 Dec '18

05 Dec '18

On Fri, Nov 30, 2018 at 03:27:21PM +0100, Stefano Garzarella wrote: > On Fri, Nov 30, 2018 at 2:06 PM Gerd Hoffmann <kraxel(a)redhat.com> wrote: > > > > Note: All these tests are done with CONFIG_DEBUG_LEVEL=0, using > > > > CONFIG_DEBUG_LEVEL=1 the boot time grows up to 24 ms, maybe we should > > > > put CONFIG_DEBUG_LEVEL=0 in the SeaBIOS configuration used in QEMU. > > > > > > I think the main seabios binary should have CONFIG_DEBUG_LEVEL=1 as it > > > helps with debug reports. I suppose an additional binary could be > > > made for those looking for the fastest possible speed. (The sole cost > > > of the debugging is the additional hardware accesses that results from > > > those debug messages.) > > > > The qemu debugcon (CONFIG_DEBUG_IO) is detecable at runtime, it returns > > 0xe9 on port reads. So we should be able to skip that too. IIRC it > > isn't *that* straightforward as seabios is initially mapped read/only so > > a simple probe-on-first-putchar, then cache the result in a variable > > doesn't work. We could probe after make_bios_writable though which > > should still avoid printing most of the messages. > > Great! I just tried the patch below and it works as you said. If I > don't have debugcon in QEMU the outs are avoided and the speed is not > bad (11.4 ms). > If I have debugcon (iobase=0x402), I can see the output. > > Do you think can be an acceptable trade-off? Makes sense to me. > index 0770c47..31c080e 100644 > --- a/src/fw/paravirt.c > +++ b/src/fw/paravirt.c > @@ -121,6 +121,10 @@ qemu_preinit(void) > kvm_detect(); > } > > + // Detect qemu debugcon > + if (inb(GET_GLOBAL(DebugOutputPort)) != QEMU_DEBUGCON_READBACK) > + DebugOutputPort = 0; As a minor quibble, this needs to be "if (CONFIG_DEBUG_IO && ...)" to ensure it gets compiled out if debug io is not in use. Also, I think it would be preferable to introduce a serial_debug_postram_preinit() method in src/hw/serialio.c that does this check. > --- a/src/hw/serialio.c > +++ b/src/hw/serialio.c > @@ -107,7 +107,7 @@ u16 DebugOutputPort VARFSEG = 0x402; > void > qemu_debug_putc(char c) > { > - if (CONFIG_DEBUG_IO && runningOnQEMU()) > + if (CONFIG_DEBUG_IO && runningOnQEMU() && GET_GLOBAL(DebugOutputPort)) > // Send character to debug port. > outb(c, GET_GLOBAL(DebugOutputPort)); > } As a minor nit, it would be preferable to invoke GET_GLOBAL(DebugOutputPort) only once in the success case. -Kevin

2 1

RFC: Add support for virt machine to Seabios
by Rob Bradford 05 Dec '18

05 Dec '18

The virt machine in the NEMU project has the ambition to be a platform that has no emulated legacy hardware. This patch series enables support for that machine type in Seabios. The impact that this has for a Seabios port is that the CMOS is not available to query details of CPUs or memory configuration; instead this patch series modifies the code that queries those details to prefer those from QEMU FW CFG over CMOS. I've tested this patch series with pc, Q35 and virt as part of our automated testing for NEMU. Rob

2 12

[PATCH] Revert "mptable: Don't describe pci-to-pci bridges."
by Liran Alon 03 Dec '18

03 Dec '18

From: Arbel Moshe <arbel.moshe(a)oracle.com> This reverts commit 3c3a3fa6522f ("mptable: Don't describe pci-to-pci bridges.”) The reverted commit removed the description of non-root PCI busses from the MPTable in claim they are not necessary. However, it seems that some guests rely on this information in order to correclty configure IOAPIC redirection-table entries for interrupts originated from PCI devices on non-root PCI busses. One such guest is "Extreme Networks OS". We observed that if this guest is setup with an E1000 NIC behind a PCI bridge, the OS wouldn't configure the IOAPIC redirection-table entry for the IRQ that the E1000 PIN is connected to. Therefore, interrupts from the NIC were not delivered to OS which caused guest to not have network connectivity. Fixes: 3c3a3fa6522f ("mptable: Don't describe pci-to-pci bridges.”) Reviewed-by: Liran Alon <liran.alon(a)oracle.com> Reviewed-by: Mark Kanda <mark.kanda(a)oracle.com> Signed-off-by: Arbel Moshe <arbel.moshe(a)oracle.com> --- src/fw/mptable.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) diff --git a/src/fw/mptable.c b/src/fw/mptable.c index 47385cc5d32d..27686369e98c 100644 --- a/src/fw/mptable.c +++ b/src/fw/mptable.c @@ -72,25 +72,30 @@ mptable_setup(void) } int entrycount = cpu - cpus; - // PCI bus + // PCI buses struct mpt_bus *buses = (void*)cpu, *bus = buses; - if (!hlist_empty(&PCIDevices)) { + int lastbus = -1; + struct pci_device *pci; + foreachpci(pci) { + int curbus = pci_bdf_to_bus(pci->bdf); + if (curbus == lastbus) + continue; + lastbus = curbus; memset(bus, 0, sizeof(*bus)); bus->type = MPT_TYPE_BUS; - bus->busid = 0; + bus->busid = curbus; memcpy(bus->bustype, "PCI ", sizeof(bus->bustype)); bus++; - entrycount++; } /* isa bus */ - int isabusid = bus - buses; + int isabusid; memset(bus, 0, sizeof(*bus)); bus->type = MPT_TYPE_BUS; - bus->busid = isabusid; + isabusid = bus->busid = lastbus + 1; memcpy(bus->bustype, "ISA ", sizeof(bus->bustype)); bus++; - entrycount++; + entrycount += bus - buses; /* ioapic */ u8 ioapic_id = BUILD_IOAPIC_ID; @@ -108,11 +113,8 @@ mptable_setup(void) int dev = -1; unsigned short pinmask = 0; - struct pci_device *pci; foreachpci(pci) { u16 bdf = pci->bdf; - if (pci_bdf_to_bus(bdf) != 0) - break; int pin = pci_config_readb(bdf, PCI_INTERRUPT_PIN); int irq = pci_config_readb(bdf, PCI_INTERRUPT_LINE); if (pin == 0) -- 2.16.1

2 5

[PATCH] mpt-scsi: Align MPT request to 8 bytes
by Liran Alon 02 Dec '18

02 Dec '18

From: Nikita Leshchenko <nikita.leshchenko(a)oracle.com> When mpt-scsi receives a SCSI message, it wraps it in a MPT request message and writes it's address to an IO port to be added to the request queue. This MPT request is allocated on the stack. Previous to this commit, the request is aligned to 4 bytes. However, VirtualBox LSI53c1030 device emulation aligns the request address to 8 bytes. Therefore, this commit change alignment of request to 8 bytes. VirtualBox source code which handles this is at Devices/Storage/DevLsiLogicSCSI.cpp. lsilogicRegisterWrite() LSILOGIC_REG_REQUEST_QUEUE handler adds the request to the queue (pRequestQueueBase). lsilogicR3Worker() reads request from pRequestQueueBase and aligns it to 8 bytes (u32RequestMessageFrameDesc & ~0x07). Reviewed-by: Liran Alon <liran.alon(a)oracle.com> Reviewed-by: Mark Kanda <mark.kanda(a)oracle.com> Signed-off-by: Nikita Leshchenko <nikita.leshchenko(a)oracle.com> --- src/hw/mpt-scsi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/hw/mpt-scsi.c b/src/hw/mpt-scsi.c index 1faede6aec6f..9966ca28054f 100644 --- a/src/hw/mpt-scsi.c +++ b/src/hw/mpt-scsi.c @@ -124,7 +124,7 @@ mpt_scsi_cmd(u32 iobase, struct disk_op_s *op, struct scsi_req { MptSCSIIORequest_t scsi_io; MptSGEntrySimple32_t sge; - } __attribute__((packed, aligned(4))) req = { + } __attribute__((packed, aligned(8))) req = { .scsi_io = { .TargetID = target, .Bus = 0, -- 2.16.1

2 4

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS December 2018