[PATCH] Recognize DVD drive during boot

During boot I had this error message sd0: TOSHIBA MQ01ABF050, 465GB w/8192kB Cache sd1: timeout waiting for ready sd1: timeout waiting for ready so the SATA DVD drive was not identified. I saw that ahci.c tries a SATA identify first, and then an ATAPI one if the first fails. But I also

Re: [PATCH] Recognize DVD drive during boot

> + if (readl(&ahci_port->cmd) & PORT_CMD_ATAPI || > + ahci_identify(ahci_host, ahci_port, port, WIN_IDENTIFY) >= 2) > + ahci_identify(ahci_host, ahci_port, port, WIN_PIDENTIFY); Thanks, that's equivalent to my code and more concise. Andrea

Re: [PATCH] Recognize DVD drive during boot

diff --git a/linux/dev/drivers/block/ahci.c b/linux/dev/drivers/block/ahci.c index b8fd9dae..0fd8d795 100644 --- a/linux/dev/drivers/block/ahci.c +++ b/linux/dev/drivers/block/ahci.c @@ -645,7 +645,7 @@ static int ahci_identify(const volatile struct ahci_host *ahci_host, const volat add

Duplicate floppy.c

Hello, gnumach has two floppy.c files and they are almost identical, as you can see from the diff. The second one seems more polished to me; also removing the first doesn't alter the build, so perhaps it is a dead file. --- linux/src/drivers/block/floppy.c2020-11-24 19:57:27.713413294 +0

master-x86_64: broken 'machine' symlink after make install-data

Hello, running "make DESTDIR=~/gnu64 install-data" on the master-x86_64 branch produces a broken "machine" symlink inside ~/gnu64/include that points to a non-existent x86_64 directory. That breaks building of mig. This is due to line "ln -s '$(systype)' '$(DESTDIR)$(include_machdir)'/machine"

gnu/hurd image: rsyslog error and likely solution

e). I wouldn't call this a priority, it's just to let you know. Warmly, Andrea Monaco

Adding a README in defunct libtreefs

gnored for all purposes." That's because I'm building some pieces of Hurd and I was stuck with the library a bit before reading that. Warmly, Andrea Monaco

Re: Adding a README in defunct libtreefs

diff --git a/libtreefs/README b/libtreefs/README new file mode 100644 index ..49477e93 --- /dev/null +++ b/libtreefs/README @@ -0,0 +1 @@ +libtreefs was never finished and is not developed anymore. +It should probably be ignored for all purposes.

Re: Adding a README in defunct libtreefs

If the code is really dead, I agree with both courses of action. Andrea

libdirmgt, possibly dead code

Hello, it seems that directory libdirmgt in hurd is dead. It is ignored by top-level Makefile, and it contains just a header file. Warmly, Andrea Monaco

[PATCH] reboot: recognize and ignore -f/--force

Hello, common boot scripts call reboot with -f or --force to force a reboot. I think our reboot can safely recognize and ignore it, to avoid breaking those scripts and make interchange of scripts easier. Andrea reboot: recognize and ignore -f/--force * hurd/sutils/reboot.c: Ditto. This wa

[PATCH] libnetfs: document return value of netfs_startup

Hello, this is trivial maybe, but the comment above netfs_startup doesn't document the return value. If I got it right, we can add it. Thanks, Andrea libnetfs: document return value of netfs_startup * libnetfs/netfs.h: Ditto. diff --git a/libnetfs/netfs.h b/libnetfs/netfs.h index d6a13b

[PATCH] console: fix listing of directory in netfs_get_dirents

led during listing, because it calls mmap with size = 0. Anyway, here's the patch. Cheers, Andrea Monaco diff --git a/console/console.c b/console/console.c index 6e83fbbc..7306b6b8 100644 --- a/console/console.c +++ b/console/console.c @@ -750,7 +750,7 @@ netfs_get_dirents (struct iouse

[PATCH] acpi: remove useless code in acpi.c

Hello, I think those lines are not meaningful: they write to an array and then never use it. Maybe they were useful for debugging. Anyway, now they can probably be removed. Warmly, Andrea Monaco diff --git a/acpi/acpi.c b/acpi/acpi.c index 63066aaf..acbf79c7 100644 --- a/acpi/acpi.c

[PATCH] acpi: handle return value of mmap consistently in acpi.c

Hello. In the first chunk, munmap is not needed, because mmap_phys_acpi_header only returns non-zero when mmap fails. In the second, a possible error condition was previously ignored. In the last chunk, I moved a common identical branch lower. Bye, Andrea Monaco diff --git a/acpi/acpi.c

acpi crashes, maybe /dev/mem is buggy?

, I see the string "XSDT" as expected; on Hurd I see zeros and then garbage. Maybe gnumach writes at that address at some point? Or it's an issue with the /dev/mem driver? I can't locate it inside gnumach. Warmly, Andrea Monaco

Re: acpi crashes, maybe /dev/mem is buggy?

> Please post the gnumach boot logs, so we can see the biosmem table. GNU Mach 1.8 biosmem: physical memory map: biosmem: 00:09e800, available biosmem: 09e800:0a, reserved biosmem: 0e:10, reserved biosmem:

Re: acpi crashes, maybe /dev/mem is buggy?

> I have now fixed both issues in gnumach master. That's great, thanks. Andrea Monaco

[PATCH] Add "comm" as a format spec to ps

which is useful when attaching gdb to a running process. Cheers, Andrea Monaco diff --git a/libps/spec.c b/libps/spec.c index c9164f9d..cbb00a3f 100644 --- a/libps/spec.c +++ b/libps/spec.c @@ -372,6 +372,17 @@ ps_get_exe (struct proc_stat *ps, char **exe_p, int *exe_len_p) } const struct ps