Hello *!

I have an Intel-based mainboard with build-in gpios that I'd like to use. However, I cannot get the kernel (5.16) to recognise the chip (NXP PCA9554). To my understanding, this should be supported with the gpio_pca953x kernel module. Upon loading this, the kernel reports no device or chip found, so all further actions via /sys/class/gpio/ or gpioinfo return nothing.

Any ideas what I might be missing here?


Regards,
Gerrit

... 6:34PM up 50 days, 23:40, 8 users, load averages: 0.52, 0.52, 0.45

--- msged/fbsd 6.3 2021-12-02
* Origin: We are the second generation (2:240/12)

Hey Gerrit!

Any ideas what I might be missing here?

$ zgrep PCA953 /proc/config.gz
CONFIG_GPIO_PCA953X=m
# CONFIG_GPIO_PCA953X_IRQ is not set
CONFIG_LEDS_PCA9532=m
# CONFIG_LEDS_PCA9532_GPIO is not set

Note that in my case they are turned off and therefore the kernel reports no device or chip found especially considering I don't have said chipset. However even if I did it should still report no device or chip found.

Turning on the PCA953X (modular or builtin) should help your cause and is the only idea I can come up with not knowing if your kernel is actually PCA953X capable.

Life is good,
Maurice

... Meotud ana wat hwær se cwealm cymeþ, þe heonon of cyþþe gewiteþ.
Only God knows where plague goes when it departs from a place.
--- GNU bash, version 5.1.16(1)-release (x86_64-lilmikii-linux-gnu)
* Origin: Little Mikey's Brain - Ladysmith BC, Canada (1:153/7001)

Hey Maurice!

$ zgrep PCA953 /proc/config.gz
CONFIG_GPIO_PCA953X=m
# CONFIG_GPIO_PCA953X_IRQ is not set
CONFIG_LEDS_PCA9532=m
# CONFIG_LEDS_PCA9532_GPIO is not set

Note that in my case they are turned off

Errrr ... only CONFIG_GPIO_PCA953X_IRQ and CONFIG_LEDS_PCA9532_GPIO are not enabled. The chipset itself is modular along with CONFIG_LEDS_PCA9532.

Getting old buddy. Mind you we got uefi boots working ... again. Not sure that is anything worth bragging about since we'd still be using legacy boots if that had been an option. :-/

Life is good,
Maurice

... Hwær cwom mearg? Hwær cwom mago? Hwær cwom maþþumgyfa?
Where is now the horse? Where the young prince? Where the treasure-giver? --- GNU bash, version 5.1.16(1)-release (x86_64-znver1-linux-gnu)
* Origin: Little Mikey's EuroPoint @ (2:280/464.113)

Hello Gerrit!

12 May 22, Gerrit Kuehn wrote to *:

I have an Intel-based mainboard with build-in gpios

Any ideas what I might be missing here?

No.

But some functions of some boards can be disabled in the bios or by jumper settings. If you're sure that the gpio are active on the hardware side then continue with software troubleshooting.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Good ${greeting_time}, Gerrit!

12 May 2022 18:33:52, you wrote to *:

I have an Intel-based mainboard with build-in gpios that I'd like
to use. However, I cannot get the kernel (5.16) to recognise the
chip (NXP PCA9554). To my understanding, this should be supported
with the gpio_pca953x kernel module.

Or not supported.

Upon loading this, the kernel reports no device or chip found, so
all further actions via /sys/class/gpio/ or gpioinfo return nothing.
Any ideas what I might be missing here?

https://www.ti.com/lit/gpn/pca9554 https://www.onsemi.com/pdf/datasheet/pca9554-d.pdf https://www.nxp.com/docs/en/data-sheet/PCA9554_9554A.pdf

Quite enough to write your own kernel module.


--
Alexey V. Vissarionov aka Gremlin from Kremlin
gremlin.ru!gremlin; +vii-cmiii-ccxxix-lxxix-xlii

... GPG: 8832FE9FA791F7968AC96E4E909DAC45EF3B1FA8 @ hkp://keys.gnupg.net
--- /bin/vi
* Origin: ::1 (2:5020/545)

Hello Maurice!

12 May 22 20:49, Maurice Kinal wrote to Gerrit Kuehn:

Turning on the PCA953X (modular or builtin) should help your cause
and is the only idea I can come up with not knowing if your kernel is actually PCA953X capable.

Well, the modules are of course loaded (I thought that was clear from what I wrote before). However, they don't find the chip which is definitely there (it's in the documentation and the board has the corresponding connectors).


Regards,
Gerrit

... 7:05AM up 51 days, 12:11, 8 users, load averages: 0.34, 0.51, 0.46

--- msged/fbsd 6.3 2021-12-02
* Origin: So come and try to tell me (2:240/12)

Hello Kai!

13 May 22 00:28, Kai Richter wrote to Gerrit Kuehn:

But some functions of some boards can be disabled in the bios or by jumper settings. If you're sure that the gpio are active on the
hardware side then continue with software troubleshooting.

There is no BIOS setting for this. Searching the web, I found one or two people reporting the same problem with the same hardware (with older kernels), but unfortunately no solution.


Regards,
Gerrit

... 7:11AM up 51 days, 12:17, 8 users, load averages: 0.48, 0.44, 0.42

--- msged/fbsd 6.3 2021-12-02
* Origin: Is serving every man (2:240/12)

Hello Maurice!

12 May 22 21:05, Maurice Kinal wrote to Maurice Kinal:

Errrr ... only CONFIG_GPIO_PCA953X_IRQ and CONFIG_LEDS_PCA9532_GPIO
are not enabled. The chipset itself is modular along with CONFIG_LEDS_PCA9532.

I don't think I'd need these. IRQ handling can be done (or not), and the LED driver is further down the userland road (and to make it even more confusing, there is leds_pca955x that would fit the chip I have). Same for i2c and sma modules.
However, my understanding is that the gpio_pca953x module should already report the chip in dmesg when being probed, and this does not happen. Everything else is useless if this first step fails.

Getting old buddy. Mind you we got uefi boots working ... again.
Not sure that is anything worth bragging about since we'd still be
using legacy boots if that had been an option. :-/

I did use legacy/csm for quite some time if available. However, meanwhile uefi works fine for me and is actually easier to set up and maintain, so I use it on all new machines now.


Regards,
Gerrit

... 9:15AM up 51 days, 14:21, 8 users, load averages: 0.25, 0.34, 0.40

--- msged/fbsd 6.3 2021-12-02
* Origin: Tall orders to fulfil (2:240/12)

Hello Alexey!

13 May 22 10:28, Alexey Vissarionov wrote to Gerrit Kuehn:

I have an Intel-based mainboard with build-in gpios that I'd like
to use. However, I cannot get the kernel (5.16) to recognise the
chip (NXP PCA9554). To my understanding, this should be supported
with the gpio_pca953x kernel module.

Or not supported.

https://github.com/torvalds/linux/blob/master/drivers/gpio/gpio-pca953x.c

{ .compatible = "nxp,pca9554", .data = OF_953X( 8, PCA_INT), },

From looking at the source, I think it should be supported by the existing module. kernelconfig has the same info:

PCA95[357]x, PCA9698, TCA64xx, and MAX7310 I/O ports

8 bits: max7310, max7315, pca6107, pca9534, pca9538, pca9554,
pca9556, pca9557, pca9574, tca6408, tca9554, xra1202

https://www.ti.com/lit/gpn/pca9554 https://www.onsemi.com/pdf/datasheet/pca9554-d.pdf https://www.nxp.com/docs/en/data-sheet/PCA9554_9554A.pdf

Quite enough to write your own kernel module.

Yeah, sure, but why bother if the driver is already there?


Regards,
Gerrit

... 10:21AM up 51 days, 15:27, 8 users, load averages: 0.36, 0.44, 0.40

--- msged/fbsd 6.3 2021-12-02
* Origin: And the pastiche we've invented (2:240/12)

Hey Gerrit!

I did use legacy/csm for quite some time if available. However,
meanwhile uefi works fine for me and is actually easier to set up
and maintain, so I use it on all new machines now.

Undertood. I am not quite at that point yet but at least I have one working with uefi. Unfortunetly I cannot get the 4k monitor into kicking in before the boot penguins appear and thus I get only 9 of 16 on the top of the screen which disappear as soon as the higher resolution kicks in. Not sure it is worth losing sleep over but I'd sure like to see the rest of the penguins which I know are there ... sort of.

Life is good,
Maurice

... Longað þonne þy læs þe him con leoþa worn.
He is less troubled by longing who knows many songs.
--- GNU bash, version 5.1.16(1)-release (x86_64-znver1-linux-gnu)
* Origin: Little Mikey's EuroPoint @ (2:280/464.113)

Hey Gerrit!

Well, the modules are of course loaded (I thought that was clear
from what I wrote before).

Probably. However I am getting older and stupider and thus often miss these sorts of clues.

That is my excuse and I am sticking to it.

Life is good,
Maurice

... Lofdædum sceal in mægþa gehwære man geþeon.
By praiseworthy deeds shall one prosper among peoples everywhere.
--- GNU bash, version 5.1.16(1)-release (x86_64-lilmikii-linux-gnu)
* Origin: Little Mikey's Brain - Ladysmith BC, Canada (1:153/7001)

Hello Maurice!

13 May 22 15:34, Maurice Kinal wrote to Gerrit Kuehn:

working with uefi. Unfortunetly I cannot get the 4k monitor into
kicking in before the boot penguins appear and thus I get only 9 of
16 on the top of the screen which disappear as soon as the higher resolution kicks in. Not sure it is worth losing sleep over but I'd
sure like to see the rest of the penguins which I know are there ...
sort of.

Sounds like a bit of a luxury problem indeed. ;)
But you're right, modern monitors are often surprisingly slow in finding signals. But that's probably with no regard to using uefi or legacy boot.


Regards,
Gerrit

... 6:28PM up 51 days, 23:34, 8 users, load averages: 0.62, 0.46, 0.43

--- msged/fbsd 6.3 2021-12-02
* Origin: We are a nation (2:240/12)

Hey Gerrit!

Sounds like a bit of a luxury problem indeed. ;)

Heh, heh. Nice for a change. It is usually el-cheapo crap problems in this neck of the woods. Not that any of this is mission critical but to quote Yosimite Sam, "I paid my five bits to see the high diving act and I am GONNA see the high diving act!"

But you're right, modern monitors are often surprisingly slow in
finding signals.

I am going to try a test kernel with all the amdgpu stuff builtin as that might fix this luxury problem without further breaking the bank.

I need a bigger budget.

But that's probably with no regard to using uefi or legacy boot.

There is a difference. With uefi I get boot penguins but not with a legacy boot, or at least not when deploying i915 or amdgpu. I have both now so I can easily check. I can't speak about amdgpu with respect to 4.x kernels but i915 was working with legacy boots. However I predict that 5.18.x kernels will find things getting noticably better. We'll soon enough see.

Life is good,
Maurice

... Treow sceal on eorle, wisdom on were.
Loyalty belongs in a warrior, wisdom in a man.
--- GNU bash, version 5.1.16(1)-release (x86_64-znver1-linux-gnu)
* Origin: Little Mikey's EuroPoint @ (2:280/464.113)

Hello Gerrit!

13 May 22, Gerrit Kuehn wrote to Maurice Kinal:

meanwhile uefi works fine for me and is actually easier to set up and maintain, so I use it on all new machines now.

Really? It took 4 or 5 days until i got my fido server up and running again. The only task i wanted to do was to move the system from a 4TB disk to a 2TB disk. To reduce the partition sizes i created the target tables from a rescue boot and rsync'd the files to their destinatitions. The partition layout was the same but with differnt size and the system failed to boot.

I'm not sure that understood anything but i looks like uefi is a mini grub within the bios. The boot device is added to the uefi boot list by uuid.

If you replace a disk on the same sata port with a newly formated disk of the same layout the boot will fail because the uuid are different.

This failure will repeat for any file based restore within grub and within the system itself because after grub the uuid of the initrd and in the /etc/fstab will not match any more.

I've no idea what the final correct solution was. Finally i was so p'd off that i dd'd the whole partitions sda1/sda2... to the new destination. I think that would kept the uuid intact and after three successful reboots i put my hands of the running systems.

I can't see the improvement there. To keep a track of the activ uuids i started to record blkid output into fstab.

The best improvement was to make use of my pxe server for the rescue boots. Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

19 May 22 18:19, Kai Richter wrote to Gerrit Kuehn:

If you replace a disk on the same sata port with a newly formated
disk of the same layout the boot will fail because the uuid are different.

You don't have to use uuids at all. Just stick to the default naming scheme. The BIOS will detect the EFI parition and start what is available there.


Regards,
Gerrit

... 6:46AM up 58 days, 11:52, 8 users, load averages: 0.87, 0.69, 0.59

--- msged/fbsd 6.3 2021-12-02
* Origin: We are a nation of innovations (2:240/12)

Hello Gerrit!

20 May 22, Gerrit Kuehn wrote to Kai Richter:

If you replace a disk on the same sata port with a newly formated
disk of the same layout the boot will fail because the uuid
are different.

You don't have to use uuids at all. Just stick to the default naming scheme.

I thought so too and i do have sda1 /boot/efi and sda3 /boot. But /boot/efi is empty.

The BIOS will detect the EFI parition and start what is available
there.

Wouldn't that add a third place to manage boot partitions?
There is the boot priority list in the bios and the boot manager grub.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

20 May 22 11:29, Kai Richter wrote to Gerrit Kuehn:

I thought so too and i do have sda1 /boot/efi and sda3 /boot. But /boot/efi is empty.

Is there an entry in fstab for it? Maybe it is just not mounted? Otherwise, have a look at the output of gdisk (or a similar tool) to locate your efi partition.

The BIOS will detect the EFI parition and start what is available
there.

Wouldn't that add a third place to manage boot partitions?
There is the boot priority list in the bios and the boot manager
grub.

The BIOS lists the bootable EFI partitions it found. Mine typically look like this:

---
~# l /boot/efi/EFI/BOOT/
-rw-rw-r-- 1 root root 139264 May 5 18:55 BOOTX64.EFI
---


This is the default naming scheme that should always work. For everything else, you might have to set up boot environment variables (usually, efibootmgr is your friend there).



Regards,
Gerrit

... 1:01PM up 58 days, 18:07, 8 users, load averages: 0.27, 0.69, 0.65

--- msged/fbsd 6.3 2021-12-02
* Origin: And the pastiche we've invented (2:240/12)

Hello Gerrit!

20 May 22, Gerrit Kuehn wrote to Kai Richter:

I thought so too and i do have sda1 /boot/efi and sda3 /boot. But

/boot/efi is empty.

Is there an entry in fstab for it?

Yes.

Maybe it is just not mounted?

Acc to mount and lsblk, no, it is mounted.

The BIOS will detect the EFI parition and start what is
available there.

Wouldn't that add a third place to manage boot partitions?
There is the boot priority list in the bios and the boot manager
grub.

The BIOS lists the bootable EFI partitions it found.

None. I think the BIOS is on csm then.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

20 May 22 23:51, Kai Richter wrote to Gerrit Kuehn:

/boot/efi is empty.

Is there an entry in fstab for it?

Yes.

Maybe it is just not mounted?

Acc to mount and lsblk, no, it is mounted.

And it's empty? So you have no EFI bootcode installed then.

The BIOS lists the bootable EFI partitions it found.

None. I think the BIOS is on csm then.

If there is no bootloader on the EFI partition as you wrote above, there is nothing the BIOS could find.


Regards,
Gerrit

... 1:26PM up 59 days, 18:32, 8 users, load averages: 0.70, 0.47, 0.39

--- msged/fbsd 6.3 2021-12-02
* Origin: We are a nation (2:240/12)

Hello Gerrit!

20 May 2022 13:01, Gerrit Kuehn wrote to Kai Richter:

I thought so too and i do have sda1 /boot/efi and sda3 /boot. But
/boot/efi is empty.

then efi is not installed with grub ?

Is there an entry in fstab for it?

uefi cant read fstab

Maybe it is just not mounted?

+1

Otherwise, have a look at the output of gdisk (or a similar tool) to locate your efi partition.

+1

The BIOS will detect the EFI parition and start what is available
there.

Wouldn't that add a third place to manage boot partitions?
There is the boot priority list in the bios and the boot manager
grub.

both can use same data, if needed

The BIOS lists the bootable EFI partitions it found. Mine typically
look like this:

-+-
~# l /boot/efi/EFI/BOOT/
-rw-rw-r-- 1 root root 139264 May 5 18:55 BOOTX64.EFI
-+-

if installed via grub

This is the default naming scheme that should always work. For
everything else, you might have to set up boot environment variables (usually, efibootmgr is your friend there).

good


Regards Benny

... too late to die young :)

--- Msged/LNX 6.1.2 (Linux/5.17.9-gentoo-dist (x86_64))
* Origin: gopher://fido.junc.eu/ (2:230/0)

Hello Benny!

22 May 22 11:18, Benny Pedersen wrote to Gerrit Kuehn:

Is there an entry in fstab for it?

uefi cant read fstab

No, but the system can, and he wants to check it.

The BIOS lists the bootable EFI partitions it found. Mine typically
look like this:

-+-
~# l /boot/efi/EFI/BOOT/
-rw-rw-r-- 1 root root 139264 May 5 18:55 BOOTX64.EFI
-+-

if installed via grub

Also when installed it manually. Even for FreeBSD. This is the default naming scheme from the EFI standard. It has nothing to do with any particular boot manager.


Regards,
Gerrit

... 3:50PM up 60 days, 20:56, 8 users, load averages: 1.47, 0.75, 0.55

--- msged/fbsd 6.3 2021-12-02
* Origin: We are a nation (2:240/12)

Hello Gerrit!

21 May 22, Gerrit Kuehn wrote to Kai Richter:

If there is no bootloader on the EFI partition as you wrote above,
there is nothing the BIOS could find.

I does find the disk. My first step was to remove the old disk and install a fresh system until ready to boot. It did and i removed the new disk to re-attach the old one. Then no boot anymore. The old disk on the same port in place of the new disk did not boot. I think that's because the bios does use some ID flags to identify the disk to boot of. I never noticed that kind of behavior on legacy systems.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hey Kai!

It did and i removed the new disk to re-attach the old one. Then
no boot anymore.

I have the identical issue between nvme and sata. On legacy it was never an issue and usually F11 ... or F12 ... on bootup would allow one to change from the default boot disk. With UEFI it has been hit and miss but usually I can get sda to boot up if nvme fails to be seen which can and does happen.

Life is good,
Maurice

... Ne byð þæt fele freond, se þe oþrum facn heleð.
He who harbours treachery against another is not a faithful friend.
--- GNU bash, version 5.1.16(1)-release (x86_64-lilmikii-linux-gnu)
* Origin: Little Mikey's Brain - Ladysmith BC, Canada (1:153/7001)

Hello Kai!

22 May 22 02:24, Kai Richter wrote to Gerrit Kuehn:

If there is no bootloader on the EFI partition as you wrote above,
there is nothing the BIOS could find.

I does find the disk.

Does it just find "the disk", or does it present it as EFI-bootable option?

new disk to re-attach the old one. Then no boot anymore. The old disk
on the same port in place of the new disk did not boot. I think
that's because the bios does use some ID flags to identify the disk
to boot of. I never noticed that kind of behavior on legacy systems.

For being EFI-bootable, the installation will need an EFI-partition with boot loader. Your description said your EFI partition was empty (no bootlader). This will not be recognised as bootable by the BIOS (of course).Maybe it was booting legacy before, and you changed the BIOS setting to EFI only?


Regards,
Gerrit

... 7:21AM up 61 days, 12:27, 8 users, load averages: 0.17, 0.37, 0.48

--- msged/fbsd 6.3 2021-12-02
* Origin: And the pastiche we've invented (2:240/12)

Hello Maurice!

22 May 22 22:44, Maurice Kinal wrote to Kai Richter:

I have the identical issue between nvme and sata. On legacy it was
never an issue and usually F11 ... or F12 ... on bootup would allow
one to change from the default boot disk.

Interestingly, this never worked reliably for me, neither with MBR nor with UEFI. The only safe way is to change BIOS boot order settings.

On the bright side, UEFI does reliably come with a shell (if everything else fails) that allows starting other systems manually.


Regards,
Gerrit

... 10:48AM up 61 days, 15:54, 8 users, load averages: 0.53, 0.62, 0.56

--- msged/fbsd 6.3 2021-12-02
* Origin: Shock to the System (2:240/12)

Hey Gerrit!

Interestingly, this never worked reliably for me, neither with MBR
nor with UEFI. The only safe way is to change BIOS boot order
settings.

That precludes that the bios recognizes the disk and presents itself as a bootable disk, which in this case is having issues with the nvme disk. Currently the only repeatable, reliable boot disk is the sata interface.

Also, while I am at it, the 16 boot penguins on a 4k monitor still requires work.

Life is good,
Maurice

... Se ðe ear gifeð and eft oftihð... bysmer he gewyrceð.
He who gives and takes it back again does a shameful thing.
--- GNU bash, version 5.1.16(1)-release (x86_64-lilmikii-linux-gnu)
* Origin: Little Mikey's Brain - Ladysmith BC, Canada (1:153/7001)

Hello Gerrit!

23 May 22, Gerrit Kuehn wrote to Kai Richter:

I does find the disk.

Does it just find "the disk", or does it present it as EFI-bootable option?

Just the disk. I'm not sure if i recognize an efi boot option. The disk is listed in the boot options by its part or model nummer.

For being EFI-bootable, the installation will need an EFI-partition
with boot loader. Your description said your EFI partition was empty
(no bootlader). This will not be recognised as bootable by the BIOS
(of course).Maybe it was booting legacy before, and you changed the
BIOS setting to EFI only?

No. With this knowledge i would say that i did boot on legacy mode always. And i still do. I'm sorry to blame efi for that because it can't be a fault of efi if efi is not used. But my perception is that this behavior change was with the uefi bios introduction.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Gerrit!

23 May 22, Gerrit Kuehn wrote to Maurice Kinal:

On the bright side, UEFI does reliably come with a shell (if
everything else fails) that allows starting other systems manually.

I still don't see the need. Doesn't have any todays operating system have a boot manager? The result is that the menu to select the starting OS is moved from the disk to the BIOS.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

24 May 22 05:39, Kai Richter wrote to Gerrit Kuehn:

On the bright side, UEFI does reliably come with a shell (if
everything else fails) that allows starting other systems manually.

I still don't see the need. Doesn't have any todays operating system
have a boot manager? The result is that the menu to select the
starting OS is moved from the disk to the BIOS.

I talked about cases where something is broken and your system doesn't boot (maybe not even the bootmanager comes up). With MBR this means reconfiguring BIOS boot order, probably getting some USB media to boot a rescue system from it etc.
With the built-in UEFI shell, you can already check your partitions and even select what to boot without having to reconfigure anything or to provide a rescue system on USB. In many cases I was able to manually boot the failed system with this and use the system itself then to fix booting.

Some manufacturers provide firmware and bios update tools on efi meanwhile. In that case, you can just drop the new firmware and the upgrade tool on a USB stick and use UEFI-shell to run it from there. In the MBR-days you would have needed to have a MSDOS-system on the USB stick in addition (and hope that the manufacturer would still provide a DOS-based update tool that is able to run on your newly bought server or workstation).


Regards,
Gerrit

... 8:00AM up 62 days, 13:06, 8 users, load averages: 0.21, 0.39, 0.43

--- msged/fbsd 6.3 2021-12-02
* Origin: Is serving every man (2:240/12)

Hello Gerrit!

24 May 22, Gerrit Kuehn wrote to Kai Richter:

The result is that the menu to select the starting OS is moved from
the disk to the BIOS.

I talked about cases where something is broken and your system doesn't boot (maybe not even the bootmanager comes up). With MBR this means reconfiguring BIOS boot order,

You have to enter the uefi shell by the same way: Dive into the BIOS and select it. If the system doesn't boot the boot override function is essential.

probably getting some USB media to boot a rescue system from it etc.
With the built-in UEFI shell, you can already check your partitions
and even select what to boot

I couldn't. The, no, my uefi shell is not self explaining. I failed to start a running system from the uefi shell. Well, ok, my fault.

without having to reconfigure anything or to provide a rescue system
on USB. In many cases I was able to manually boot the failed system
with this and use the system itself then to fix booting.

Some manufacturers provide firmware and bios update tools on efi meanwhile. In that case, you can just drop the new firmware and the upgrade tool on a USB stick and use UEFI-shell to run it from there.

So i have to wait until my BIOS company updates it's tools and i can save them within the BIOS. Back in the old days i had my OS/2 bootmanager and rescue system on a small partition.

In the MBR-days you would have needed to have a MSDOS-system on the
USB stick in addition

Why DOS? We need to kickstart the boot sequence.
USB stick in addition? I see an EFI partition on my disk in addition.

I've done 50% reading of an efi turorial now and it does remind me to grub.

(and hope that the manufacturer would still provide a DOS-based
update tool that is able to run on your newly bought server or workstation).

I'v seen an EFI editor which my bios doesn't have. Now i've to check if anyone provides this efi software. The default path in my uefi shell does have something like fs[01234]:/efi/tools. The design concept is to add tools from a maintenance/rescue partition like we did before from the usb stick or from a /boot partition.

Maybe the only thing i need is to get used to it...

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

27 May 22 13:13, Kai Richter wrote to Gerrit Kuehn:

I talked about cases where something is broken and your system
doesn't boot (maybe not even the bootmanager comes up). With MBR this
means reconfiguring BIOS boot order,

You have to enter the uefi shell by the same way: Dive into the BIOS
and select it. If the system doesn't boot the boot override function
is essential.

No, the default config of a BIOS usually is to come up with EFI shell if everything else fails.

I couldn't. The, no, my uefi shell is not self explaining. I failed
to start a running system from the uefi shell. Well, ok, my fault.

Yes, that's the part that took me some time, too. ;)
However, the basic things are not that hard once you find out how it works.

Some manufacturers provide firmware and bios update tools on efi
meanwhile. In that case, you can just drop the new firmware and the
upgrade tool on a USB stick and use UEFI-shell to run it from there.

So i have to wait until my BIOS company updates it's tools and i can
save them within the BIOS.

?
Sorry, you lost me there.

In the MBR-days you would have needed to have a MSDOS-system on the
USB stick in addition

Why DOS? We need to kickstart the boot sequence.

I was talking about updating BIOS or firmware in your system here, not getting a broken system online again. Updating usually involves some kind of Windows software. As I don't use Windows, this doesn't help me. The fallback option usually is something DOS-based (if it exists), but this requires some extra steps. Having a EFI-based update tool is less of a hassle.

I'v seen an EFI editor which my bios doesn't have. Now i've to check
if anyone provides this efi software.

Which software are you looking for? Most things can be set up with efibootmgr under Linux (or FreeBSD).


Regards,
Gerrit

... 4:18PM up 65 days, 21:24, 8 users, load averages: 0.59, 0.50, 0.51

--- msged/fbsd 6.3 2021-12-02
* Origin: We are the second generation (2:240/12)

Hello Gerrit!

27 May 22, Gerrit Kuehn wrote to Kai Richter:

Some manufacturers provide firmware and bios update tools on efi
meanwhile. In that case, you can just drop the new firmware and
the upgrade tool on a USB stick and use UEFI-shell to run it
from there.

So i have to wait until my BIOS company updates it's tools and i
can save them within the BIOS.

Sorry, you lost me there.

Ups, i deleted that part. My point was that we have wait for the manufacturers to provide firmware. This does happen out of our control anyway. No matter if uefi or csm - in both cases we have to rely to the manufacturer. My bios does have a flashtool inclueded but i have to format an usb storage device to fat32 and copy the bios to it. I don't see much difference if i dd an image or copy the file. I see the efi partition as somthing like a permanent usb stick.

In the MBR-days you would have needed to have a MSDOS-system on
the USB stick in addition

Why DOS? We need to kickstart the boot sequence.

I was talking about updating BIOS or firmware in your system here, not getting a broken system online again.

Same task, we need to kickstart the boot sequence. The differnce between uefi and csm is the storage location of the whatever starting software.

Updating usually involves some kind of Windows software.

My last one extracted a windows disk imager and copied the image to floppy. Those are self-extraction compressed exes sometimes that can be unzipped or unarj'd to get the real image.

I'v seen an EFI editor which my bios doesn't have. Now i've to
check if anyone provides this efi software.

Which software are you looking for?

None. It looked like ee. uefi adds a layer of complexity. If i want to use uefi like a rescue system i can't simply use my known tools from my disk. I think i will stick on my rescue stick in case of trouble because the tools i'm used to are available.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

30 May 22 02:12, Kai Richter wrote to Gerrit Kuehn:

rely to the manufacturer. My bios does have a flashtool inclueded but
i have to format an usb storage device to fat32 and copy the bios to
it. I don't see much difference if i dd an image or copy the file. I
see the efi partition as somthing like a permanent usb stick.

The difference is that without UEFI you need DOS on the USB stick, not just your BIOS flash tool and image. Ever tried booting modern server-hardware under DOS and running BIOS-update tools from that?

Why DOS? We need to kickstart the boot sequence.

I was talking about updating BIOS or firmware in your system here,
not getting a broken system online again.

Same task, we need to kickstart the boot sequence.

No, we need to come up with an environment that is able to run the flash tool the manufacturer provides. If this tool is for EFI shell, this is easy. Getting a DOS tool to run under DOS booted from a USB stick... your mileage may vary.

check if anyone provides this efi software.

Which software are you looking for?

None. It looked like ee.

You lost me again.

uefi adds a layer of complexity. If i want
to use uefi like a rescue system i can't simply use my known tools
from my disk. I think i will stick on my rescue stick in case of
trouble because the tools i'm used to are available.

I didn't mean to say that EFI might fully replace your rescue disk (whatever it looks like), it certainly doesn't. But in some at least for me quite common cases where you just want to get the system booted it serves that purpose and relieves me from having to boot the rescue disk.


Regards,
Gerrit

... 9:28AM up 70 days, 14:34, 8 users, load averages: 0.42, 0.48, 0.50

--- msged/fbsd 6.3 2021-12-02
* Origin: America, America the western dream is gone (2:240/12)

Hello Gerrit!

01 Jun 22, Gerrit Kuehn wrote to Kai Richter:

The difference is that without UEFI you need DOS on the USB stick, not just your BIOS flash tool and image. Ever tried booting modern server-hardware under DOS and running BIOS-update tools from that?

No, i use PXE images for that. Doesn't have a modern server IPMI and does the flash via ipmi bios image upload?

Same task, we need to kickstart the boot sequence.

No, we need to come up with an environment that is able to run the
flash tool the manufacturer provides.

No matter what OS runs for those tools - we still need to kickstart them.

If this tool is for EFI shell, this is easy. Getting a DOS tool to
run under DOS booted from a USB stick... your mileage may vary.

This is the responsiblity of the hardware manufacturer. Btw, let's look on the uefi shell. It can read partitions and filesystems and load/start software from. Do you remember toms root boot on a floppy disk? Basically it does the same. The uefi shell is not a shell of the bios - it's an operating system. If the flash tool is efi software the efi shell OS could be placed on the USB stick too. And again, the only job of the BIOS (remember it's BASIC IO) is to kickstart the boot sequence for an OS.

I didn't mean to say that EFI might fully replace your rescue disk (whatever it looks like), it certainly doesn't. But in some at least
for me quite common cases where you just want to get the system booted

Which common cases on uefi systems do break the boot process?

it serves that purpose and relieves me from having to boot the rescue disk.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

02 Jun 2022 23:43, Kai Richter wrote to Gerrit Kuehn:

No, i use PXE images for that. Doesn't have a modern server IPMI and
does the flash via ipmi bios image upload?

my bios have ipmi, but imho not supported image upload to use it ?


Regards Benny

... too late to die young :)

--- Msged/LNX 6.1.2 (Linux/5.17.12-gentoo-dist (x86_64))
* Origin: gopher://fido.junc.eu/ (2:230/0)

Hello Benny!

03 Jun 22, Benny Pedersen wrote to Kai Richter:

No, i use PXE images for that. Doesn't have a modern server IPMI
and does the flash via ipmi bios image upload?

my bios have ipmi, but imho not supported image upload to use it ?

I'm sorry to say that i can't take a closer look now. I'm sure there was something like upload but i'm unsure if it was upload for the ipmi image or the bios image. I got a working PXE boot before i got an ipmi board so i continued with PXE and did't dive into the details of ipmi standards.

On the other hand i think ipmi would be incomplete without bios update function because it's designed for remote system administration. As far as i remember it can load the boot image and at least that way it can load the flash disk with OS and flash tools.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai!

04 Jun 22 06:53, Kai Richter wrote to Benny Pedersen:

I'm sorry to say that i can't take a closer look now. I'm sure there
was something like upload but i'm unsure if it was upload for the
ipmi image or the bios image.

Depends. Supermicro, e.g., offers both. But only IPMI firmware updates are for free, For being able to update the mainboard's BIOS via IPMI, you have to buy a license key (or hack it). Updating firmware or bios of plugin cards like storage controllers is a completely different kettle of fish, again.


Regards,
Gerrit

... 9:53PM up 74 days, 2:59, 8 users, load averages: 0.71, 0.58, 0.57

--- msged/fbsd 6.3 2021-12-02
* Origin: We are a nation (2:240/12)

Maurice! Hello. Prism was down for a long time but I tried logging in today and it worked! Nice to have my usual
Fido Gateway available.


I was looking around to see the current state of ttylinux and see it's not really active anymore, and found that even some of the downloads were no longer available.

I had an idea that I thought might be good, and after find a couple isos, I pushed them to my docker account. I felt like people may still be interested in the future, at least for archiving/historical purposes. What do you think?

https://hub.docker.com/repository/docker/andy5995/ttylinux/general

Anyway, how have you been? Me and a buddy have been working on doing a Linux From Scratch installation. I had done it about a decade ago, but he's never, so I thought it would be fun to do with him. We've booted into it and now
just have to get his wifi working. Fun!

--
-Andy


... Computer Hacker wanted. Must have own axe.
-+- MultiMail/Linux v0.52-c9c24da

--- BBBS/Li6 v4.10 Toy-6
* Origin: Prism bbs (1:261/38)

Anyway, how have you been? Me and a buddy have been working on doing a Linux From Scratch installation. I had done it about a decade ago, but he's never, so I thought it would be fun to do with him. We've booted into it and now just have to get his wifi working. Fun!

Sounds like you need network manager & friends.

I just did this myself on my desktop, it just needed dhcpcd so it was fairly simple. Network manager might be more involved, I'll have to go down that road too when I get to my laptop.

Have fun!

--- BBBS/Li6 v4.10 Toy-6
* Origin: The Rusty MailBox - Penticton, BC Canada (1:153/757)

Hey Andrew!

Prism was down for a long time but I tried logging in today and
it worked!

Fantastic news. I'll give it a try later. Thanks for the heads up.

I was looking around to see the current state of ttylinux and see
it's not really active anymore

I haven't checked it out for awhile but last time I did it looked to be heading in that direction. I was toying with the idea of doing something similar for a bbs, perhaps a virtual one?

I felt like people may still be interested in the future, at
least for archiving/historical purposes. What do you think?

Excellent idea. I would probably check that out.

Life is good,
Maurice

o- o- -o -o -o o- o- -o o- o- o- o- -o -o -o o- /) /) (\ (\ (\ /) /) (\ /) /) /) /) (\ (\ (\ /) ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ... Fidonet 4K - You load sixteen penguins and what do you get?
--- GNU bash, version 5.2.15(1)-release (x86_64-pc-linux-gnu)
* Origin: One of us @ (1:153/7001.2989)

Alan Ianson wrote to Andrew Alt <=-

Anyway, how have you been? Me and a buddy have been working on doing a Linux From Scratch installation. I had done it about a decade ago, but he's never, so I thought it would be fun to do with him. We've booted into it and now just have to get his wifi working. Fun!

Sounds like you need network manager & friends.

We got it working today with wpa_supplicant and dhclient! :)


--
-Andy


... Internal Error: The system has been taken over by sheep at line 19960
-+- MultiMail/Linux v0.52-c9c24da

--- BBBS/Li6 v4.10 Toy-6
* Origin: Prism bbs (1:261/38)

Maurice Kinal wrote to Andrew Alt <=-

I haven't checked it out for awhile but last time I did it looked to be heading in that direction. I was toying with the idea of doing
something similar for a bbs, perhaps a virtual one?

I participated in a recent discussion on Reddit. Someone asked about running Synchronet BBS in a docker container.

Another person said ze got Mystic BBS running in a docker with dosemu.

Also someone brought up making an appimage build of synch bbs. ;)

https://www.reddit.com/r/bbs/comments/zwjyas/dockerized/


--
-Andy


... "42? 7 and a half million years and all you can come up with is 42?!"
-+- MultiMail/Linux v0.52-c9c24da

--- BBBS/Li6 v4.10 Toy-6
* Origin: Prism bbs (1:261/38)

Sounds like you need network manager & friends.

We got it working today with wpa_supplicant and dhclient! :)

Good stuff. LFS is (one of) my favorite things these days.

--- BBBS/Li6 v4.10 Toy-6
* Origin: The Rusty MailBox - Penticton, BC Canada (1:153/757)

Hey Andrew!

Someone asked about running Synchronet BBS in a docker container.

I was thinking more along the line of a standard ansi-type bbs except in my case it would serve up the 'linux' terminal which in my case also happens to have framebuffer graphics available to it. That way it potentially could serve up pdf and the such. I think a stock ttylinux could easily handle something like this although I'd have to work out the graphical output ... also utf-8 will require a bit of jiggery-pokery.

Life is good,
Maurice

o- o- -o o- -o o- o- -o o- o- -o -o o- o- o- -o /) /) (\ /) (\ /) /) (\ /) /) (\ (\ /) /) /) (\ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ... Fidonet 4K - You load sixteen penguins and what do you get?
--- GNU bash, version 5.2.15(1)-release (x86_64-pc-linux-gnu)
* Origin: One of us @ (1:153/7001.2989)

Maurice Kinal wrote to Andrew Alt <=-

Hey Andrew!

Someone asked about running Synchronet BBS in a docker container.

I was thinking more along the line of a standard ansi-type bbs except
in my case it would serve up the 'linux' terminal which in my case also happens to have framebuffer graphics available to it. That way it potentially could serve up pdf and the such. I think a stock ttylinux could easily handle something like this although I'd have to work out
the graphical output ... also utf-8 will require a bit of
jiggery-pokery.

Sounds interesting. Last week I started updating ttylinux (https://github.com/andy5995/ttylinux/pull/1/files) but it's doubtful I'll continue with it on my own.


--
-Andy
Website: https://andy5995.github.io


... I am not Paul.
-+- MultiMail/Linux v0.52-c9c24da

--- BBBS/Li6 v4.10 Toy-6
* Origin: Prism bbs (1:261/38)

Hey Andrew!

Last week I started updating ttylinux ... but it's doubtful I'll
continue with it on my own.

It is far too much of a commitment for this old man. These days I am sticking to playing and helping out where I can. As for lfs, I doubt I could quit at this stage of the game. I got hooked around the same time I first started playing with ttylinux. For me they almost became the same thing.

Life is good,
Maurice

-o o- o- -o
(\ /) /) (\
^^ ^^ ^^ ^^
... I've got squirrels in my pants!
--- GNU bash, version 5.2.15(1)-release (x86_64-pc-linux-gnu)
* Origin: Little Mikey's Brain - Ladysmith BC, Canada (1:153/7001)

Maurice Kinal wrote to Andrew Alt <=-

Hey Andrew!

Last week I started updating ttylinux ... but it's doubtful I'll
continue with it on my own.

It is far too much of a commitment for this old man. These days I am sticking to playing and helping out where I can. As for lfs, I doubt I could quit at this stage of the game. I got hooked around the same
time I first started playing with ttylinux. For me they almost became
the same thing.

It's good to know where you're at. :)


--
-Andy
Website: https://andy5995.github.io


-+- MultiMail/Linux v0.52-c9c24da

--- BBBS/Li6 v4.10 Toy-6
* Origin: Prism bbs (1:261/38)

Hey Andrew!

It's good to know where you're at. :)

I suppose it would be. :-/

Life is good,
Maurice

o- o- -o o-
/) /) (\ /)
^^ ^^ ^^ ^^
... Hægl byþ hwitust corna; hwyrft hit of heofones lyfte.
Hail is the whitest of grains; it whirls down from the heavens.
--- GNU bash, version 5.2.15(1)-release (x86_64-pc-linux-gnu)
* Origin: Little Mikey's Brain - Ladysmith BC, Canada (1:153/7001)