OpenBIOS

openbios@openbios.org

4289 discussions

[PATCH v3 0/3] esp: fixups to correctly terminate successful SCSI commands
by Mark Cave-Ayland 22 Sep '24

22 Sep '24

This series is an extension of the v2 patch with a couple of modifications to clear the FIFO at the start of the command, and to poll the ESP interrupt register to detect the end of the transfer. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> v3: - Clear FIFO before issuing ESP command - Poll ESP interrupt register to detect end of transfer v2: - Fix typo in commit message spotted by Zoltan Mark Cave-Ayland (3): esp: clear FIFO before issuing ESP command esp: use ESP rather than DMA interrupt to detect end of transfer esp: correctly terminate successful SCSI commands drivers/esp.c | 37 ++++++++++++++++++++++++------------- 1 file changed, 24 insertions(+), 13 deletions(-) -- 2.39.2

1 4

[PATCH v2] README: change reference from LinuxBIOS to coreboot
by Mark Cave-Ayland 22 Sep '24

22 Sep '24

The linuxbios.org website does not exist anymore, so point the reader towards coreboot.org instead. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Resolves: https://github.com/openbios/openbios/issues/17 Reviewed-by: Philippe Mathieu-Daudé <philmd(a)linaro.org> Tested-by: Philippe Mathieu-Daudé <philmd(a)linaro.org> --- README | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) v2: - Add R-B and T-B tags from Phil - Replace Coreboot with coreboot as indicated by Segher diff --git a/README b/README index 9f4b705..bf7684a 100644 --- a/README +++ b/README @@ -17,8 +17,8 @@ kernel and three interfaces for user interaction, device initialization and client (operating system) control. While far not all possible applications of OpenBIOS are implemented yet, -a lot of functionality is already there. OpenBIOS can be used to enhance -LinuxBIOS (http://www.linuxbios.org), or be booted from any multiboot +a lot of functionality is already there. OpenBIOS can be used as a payload +for coreboot (http://www.coreboot.org), or be booted from any multiboot capable bootloader to bring Open Firmware to your machine. OpenBIOS can also be used when an operating system is already running. It provides the needed OpenFirmware functionality to MOL (MacOnLinux) to boot MacOS -- 2.39.2

1 1

[PATCH] cuda: fix len calculation in cuda_adb_req
by Amadeusz Sławiński 22 Sep '24

22 Sep '24

When advancing buffer by 1, len should be shorthened by appropriate amount, instead of setting it to -1. Found with cppcheck. Signed-off-by: Amadeusz Sławiński <amade(a)asmblr.net> --- drivers/cuda.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/cuda.c b/drivers/cuda.c index 86d2ddb..93a935a 100644 --- a/drivers/cuda.c +++ b/drivers/cuda.c @@ -162,7 +162,7 @@ static int cuda_adb_req (void *host, const uint8_t *snd_buf, int len, } } else { pos = buffer + 1; - len = -1; + len -= 1; } memcpy(rcv_buf, pos, len); -- 2.44.0

2 2

[PATCH v2] esp: correctly terminate successful SCSI commands
by Mark Cave-Ayland 11 Sep '24

11 Sep '24

An upcoming series for QEMU's ESP emulation will check that ESP commands are only issued according to their valid states as listed in the datasheet, and otherwise generate an illegal command interrupt. Currently if a SCSI command is expected to return no data or transfers data successfully, the ASC is not returned to its disconnected state. This means that any subsequent ESP initiator commands should be rejected until the SCSI bus transaction is terminated. Update the ESP driver so that if a SCSI command is successful then the ICCS and MSGACC commands are issued by the host to complete the SCSI transaction. This ensures that the ASC is returned to the disconnected state which allows the ESP to accept subsequent SCSI commands without generating an illegal command interrupt. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> --- drivers/esp.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) v2: - Fix typo in commit message spotted by Zoltan diff --git a/drivers/esp.c b/drivers/esp.c index 0880ab2..b1b10f4 100644 --- a/drivers/esp.c +++ b/drivers/esp.c @@ -113,7 +113,8 @@ do_command(esp_private_t *esp, sd_private_t *sd, int cmdlen, int replylen) return status; } if (replylen == 0) { - return 0; + /* No reply expected, ICCS and MSGACC to complete */ + goto done; } /* Target went to status phase instead of data phase? */ if ((status & ESP_STAT_PMASK) == ESP_STATP) { @@ -139,10 +140,18 @@ do_command(esp_private_t *esp, sd_private_t *sd, int cmdlen, int replylen) DPRINTF("do_command_reply: status 0x%x\n", status); - if ((status & ESP_STAT_TCNT) != ESP_STAT_TCNT) + if ((status & ESP_STAT_TCNT) != ESP_STAT_TCNT) { return status; - else - return 0; // OK + } + +done: + // ICCS and MSGACC sequence to complete + esp->ll->regs[ESP_CMD] = ESP_CMD_ICCSEQ; + (void)esp->ll->regs[ESP_FDATA]; // Ignore 1st byte for now + (void)esp->ll->regs[ESP_FDATA]; // Ignore 2nd byte for now + esp->ll->regs[ESP_CMD] = ESP_CMD_MOK; + + return 0; // OK } // offset is in sectors -- 2.39.2

1 1

[PATCH] esp: correctly terminate successful SCSI commands
by Mark Cave-Ayland 08 Sep '24

08 Sep '24

An upcoming series for QEMU's ESP emulation will check that ESP commands are only issued according to their valid states as listed in the datasheet, and otherwise generate an illegal command interrupt. Currently if a SCSI command is expected to return no data or transfers data successfully, the ASC is not returned to its disconnected state. This means that any subsequent ESP initiator commands should be rejected until the the SCSI bus transaction is terminated. Update the ESP driver so that if a SCSI command is successful then the ICCS and MSGACC commands are issued by the host to complete the SCSI transaction. This ensures that the ASC is returned to the disconnected state which allows the ESP to accept subsequent SCSI commands without generating an illegal command interrupt. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> --- drivers/esp.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/drivers/esp.c b/drivers/esp.c index 0880ab2..b1b10f4 100644 --- a/drivers/esp.c +++ b/drivers/esp.c @@ -113,7 +113,8 @@ do_command(esp_private_t *esp, sd_private_t *sd, int cmdlen, int replylen) return status; } if (replylen == 0) { - return 0; + /* No reply expected, ICCS and MSGACC to complete */ + goto done; } /* Target went to status phase instead of data phase? */ if ((status & ESP_STAT_PMASK) == ESP_STATP) { @@ -139,10 +140,18 @@ do_command(esp_private_t *esp, sd_private_t *sd, int cmdlen, int replylen) DPRINTF("do_command_reply: status 0x%x\n", status); - if ((status & ESP_STAT_TCNT) != ESP_STAT_TCNT) + if ((status & ESP_STAT_TCNT) != ESP_STAT_TCNT) { return status; - else - return 0; // OK + } + +done: + // ICCS and MSGACC sequence to complete + esp->ll->regs[ESP_CMD] = ESP_CMD_ICCSEQ; + (void)esp->ll->regs[ESP_FDATA]; // Ignore 1st byte for now + (void)esp->ll->regs[ESP_FDATA]; // Ignore 2nd byte for now + esp->ll->regs[ESP_CMD] = ESP_CMD_MOK; + + return 0; // OK } // offset is in sectors -- 2.39.2

2 3

[PATCH] README: change reference from LinuxBIOS to Coreboot
by Mark Cave-Ayland 06 Sep '24

06 Sep '24

The linuxbios.org website does not exist anymore, so point the reader towards coreboot.org instead. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Resolves: https://github.com/openbios/openbios/issues/17 --- README | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README b/README index 9f4b705..7e299b0 100644 --- a/README +++ b/README @@ -17,8 +17,8 @@ kernel and three interfaces for user interaction, device initialization and client (operating system) control. While far not all possible applications of OpenBIOS are implemented yet, -a lot of functionality is already there. OpenBIOS can be used to enhance -LinuxBIOS (http://www.linuxbios.org), or be booted from any multiboot +a lot of functionality is already there. OpenBIOS can be used as a payload +for Coreboot (http://www.coreboot.org), or be booted from any multiboot capable bootloader to bring Open Firmware to your machine. OpenBIOS can also be used when an operating system is already running. It provides the needed OpenFirmware functionality to MOL (MacOnLinux) to boot MacOS -- 2.39.2

3 3

[PATCH 0/6] Fixes for Forth PCI config access words and implementation
by Mark Cave-Ayland 30 Jul '24

30 Jul '24

This series provides a set of fixes required to allow Forth PCI config access words to work correctly on both 32-bit and 64-bit platforms, along with Zoltan's original v4 patch. There are 2 main fixes included: - Implement and convert PCI encode-unit/decode-unit to static methods - Fix my-address and my-unit on 64-bit platforms With this series applied it is possible to execute the Forth PCI config words as documented in various drivers as below: PPC (32-bit): 0 > showstack ok 0 > " /pci/QEMU,VGA" open-dev to my-self ok 0 > my-space " config-w@" $call-parent u. 1234 ok 0 > my-space ok 800 < 1 > my-address ok 800 0 0 < 3 > SPARC64 (64-bit, including across PCI bridge): 0 > showstack ok 0 > " /pci/pci@1,1/QEMU,VGA" open-dev to my-self ok 0 > my-space " config-w@" $call-parent u. 1234 ok 0 > my-space ok 11000 < 1 > my-address ok 11000 0 0 < 3 > Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> BALATON Zoltan (1): drivers/pci: Implement config access words Mark Cave-Ayland (5): libopenbios/clib.fs: add new is-2arg-cfunc for calling into C libopenbios: introduce BIND_NODE_STATIC_METHOD and supporting functions drivers/pci.c: convert encode-unit and decode-unit words to static methods package.fs: fix my-address on 64-bit systems package.fs: fix my-unit on 64-bit systems drivers/pci.c | 115 +++++++++++++++++++++++++++++++-- forth/device/package.fs | 8 +-- include/libopenbios/bindings.h | 7 ++ libopenbios/bindings.c | 26 ++++++++ libopenbios/clib.fs | 10 +++ 5 files changed, 155 insertions(+), 11 deletions(-) -- 2.30.2

2 14

[PATCH] kernel: fix spelling of endianness
by Mark Cave-Ayland 30 Apr '24

30 Apr '24

This is based upon a similar patch currently being carried in the QEMU Debian package sources. Reported-by: Michael Tokarev <mjt(a)tls.msk.ru> Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> --- kernel/bootstrap.c | 4 ++-- kernel/dict.c | 2 +- kernel/include/dict.h | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-) diff --git a/kernel/bootstrap.c b/kernel/bootstrap.c index 9bc94bb..8ad3efe 100644 --- a/kernel/bootstrap.c +++ b/kernel/bootstrap.c @@ -171,9 +171,9 @@ static void write_dictionary(const char *filename) .version = 2, .cellsize = sizeof(ucell), #ifdef CONFIG_BIG_ENDIAN - .endianess = -1, + .endianness = -1, #else - .endianess = 0, + .endianness = 0, #endif .checksum = 0, .compression = 0, diff --git a/kernel/dict.c b/kernel/dict.c index 0986cb1..51afc67 100644 --- a/kernel/dict.c +++ b/kernel/dict.c @@ -252,7 +252,7 @@ void dump_header(dictionary_header_t *header) printk("OpenBIOS dictionary:\n"); printk(" version: %d\n", header->version); printk(" cellsize: %d\n", header->cellsize); - printk(" endianess: %s\n", header->endianess?"big":"little"); + printk(" endianness: %s\n", header->endianness?"big":"little"); printk(" compression: %s\n", header->compression?"yes":"no"); printk(" relocation: %s\n", header->relocation?"yes":"no"); printk(" checksum: %08x\n", target_long(header->checksum)); diff --git a/kernel/include/dict.h b/kernel/include/dict.h index 749fd6f..1e735fa 100644 --- a/kernel/include/dict.h +++ b/kernel/include/dict.h @@ -27,7 +27,7 @@ typedef struct dictionary_header { char signature[8]; u8 version; u8 cellsize; - u8 endianess; + u8 endianness; u8 compression; u8 relocation; u8 reserved[3]; -- 2.39.2

1 1

Re: [PATCH 4/7] drivers/usb.c: rework descriptor read logic in get_descriptor()
by Mark Cave-Ayland 25 Apr '24

25 Apr '24

On 24/04/2024 15:48, Philippe Mathieu-Daudé wrote: > Hi Mark, > > On 22/4/24 09:54, Mark Cave-Ayland wrote: >> When compiling with gcc 12 drivers/usb.c generates the following error: >> >> /root/drivers/usb.c: In function 'get_descriptor': >> /root/drivers/usb.c:200:23: error: array subscript 'device_descriptor_t[0]' is >> partly outside array bounds of 'u8[8]' {aka 'unsigned char[8]'} [-Werror=array-bounds] >> 200 | if (dd->bMaxPacketSize0 != 0) >> | ^~ >> /root/drivers/usb.c:181:12: note: object 'buf' of size 8 >> 181 | u8 buf[8]; >> | ^~~ >> cc1: all warnings being treated as errors >> make[1]: *** [rules.mak:229: target/drivers/usb.o] Error 1 >> make[1]: Leaving directory '/root/obj-ppc' >> >> The compiler is complaining that a device_descriptor_t pointer cast to the 8 byte >> buffer >> means that a pointer dereference to some members of the device_descriptor_t could go >> beyond the end of the 8 byte buffer. Whilst this sounds as if it should be allowed, >> some >> searching suggests that the behaviour of such a cast is undefined. >> >> Rework get_descriptor() so that an entire device_descriptor_t is used to store the >> first >> 8 bytes of the incoming descriptor, and update the buf references to use the >> equivalent >> device_descriptor_t fields instead. >> >> Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> >> --- >> drivers/usb.c | 17 ++++++++--------- >> 1 file changed, 8 insertions(+), 9 deletions(-) >> >> diff --git a/drivers/usb.c b/drivers/usb.c >> index 88b7580..d0e1ae5 100644 >> --- a/drivers/usb.c >> +++ b/drivers/usb.c >> @@ -178,7 +178,7 @@ u8 * >> get_descriptor (usbdev_t *dev, unsigned char bmRequestType, int descType, >> int descIdx, int langID) >> { >> - u8 buf[8]; >> + device_descriptor_t dd; >> u8 *result; >> dev_req_t dr; >> int size; >> @@ -189,24 +189,23 @@ get_descriptor (usbdev_t *dev, unsigned char bmRequestType, >> int descType, >> dr.wValue = __cpu_to_le16((descType << 8) | descIdx); >> dr.wIndex = __cpu_to_le16(langID); >> dr.wLength = __cpu_to_le16(8); >> - if (dev->controller->control (dev, IN, sizeof (dr), &dr, 8, buf)) { >> + if (dev->controller->control (dev, IN, sizeof (dr), &dr, 8, (u8 *)&dd)) { > > Still using dalen=8 seems odd. You only fill the first 8 bytes of dd, > leaving the rest uninitialized. Looking at which fields are used, it > is OK, but hard to review and possibly bugprone if this function were > modified in the future. > > Why not use sizeof (dd) instead? That still fits into a page, so > performance overhead is unlikely. > > My 2 cents, otherwise LGTM. Thanks! I did consider whether this would matter, but given that only the first 8 bytes were accessed it didn't seem to matter where the extra junk at the end of device_descriptor_t came from, so I kept it simple. The corresponding file upstream is very different from what is in OpenBIOS, so possibly this could be made clearer with an update, but that requires quite a bit more work :/ >> usb_debug ("getting descriptor size (type %x) failed\n", >> descType); >> } >> if (descType == 1) { >> - device_descriptor_t *dd = (device_descriptor_t *) buf; >> - usb_debug ("maxPacketSize0: %x\n", dd->bMaxPacketSize0); >> - if (dd->bMaxPacketSize0 != 0) >> - dev->endpoints[0].maxpacketsize = dd->bMaxPacketSize0; >> + usb_debug ("maxPacketSize0: %x\n", dd.bMaxPacketSize0); >> + if (dd.bMaxPacketSize0 != 0) >> + dev->endpoints[0].maxpacketsize = dd.bMaxPacketSize0; >> } >> /* special case for configuration descriptors: they carry all their >> subsequent descriptors with them, and keep the entire size at a >> different location */ >> - size = buf[0]; >> - if (buf[1] == 2) { >> - int realsize = __le16_to_cpu(((unsigned short *) (buf + 2))[0]); >> + size = dd.bLength; >> + if (dd.bDescriptorType == 2) { >> + int realsize = __le16_to_cpu(dd.bcdUSB); >> size = realsize; >> } >> result = malloc (size); ATB, Mark.

1 0

[PATCH 0/7] Build updates and improvements
by Mark Cave-Ayland 24 Apr '24

24 Apr '24

This series is based upon Amadeusz Sławiński's recent series at https://mail.coreboot.org/hyperkitty/list/openbios@openbios.org/thread/OLS2… and updates the code to fix build issues on gcc 12 (as used by Debian stable), whilst also correcting the compiler flags so that the binaries are generated correctly. Finally the github workflows are updated to use the Debian cross-compilers, and the GitHub release action changed to use an actively maintained fork. [MCA: switching to the Debian cross-compilers substantially lowers the barrier to entry for people setting up their own compile environments and also brings OpenBIOS in line with QEMU's own CI cross-compiler containers]. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Mark Cave-Ayland (7): .github/workflows: update to latest action releases arch/ppc/qemu/init.c: work around GCC bug for ppc64_patch_handlers() packages/pc-parts.c: fix bug in extended partition logic drivers/usb.c: rework descriptor read logic in get_descriptor() switch-arch: disable generation of position independent code (-fno-pic) docker: switch from kernel crosstools compilers to Debian cross-compilers .github/workflows: migrate release action to crowbarmaster/GH-Automatic-Releases .github/workflows/build-openbios-builder.yml | 8 ++++---- .github/workflows/main.yml | 8 ++++---- .github/workflows/release.yml | 8 ++++---- arch/ppc/qemu/init.c | 12 ++++++++++++ config/scripts/switch-arch | 10 +++++----- docker/Dockerfile.builder | 18 ++---------------- drivers/usb.c | 17 ++++++++--------- packages/pc-parts.c | 7 +++++-- 8 files changed, 44 insertions(+), 44 deletions(-) -- 2.39.2

2 11

Jump to page: