Hi, Since the last system update I have been getting TPM errors on every entry in my grub menu when I do a warm restart from F38, and this includes the menu entry for the kernel I used to boot into F38. I am getting multiple occurrences of the following message: error: ../../grub-core/commands/efi/tpm.c:150:unknown TPM error.
I am also getting the message: error: ../../grub-core/net/net.c:1552:disk 'hd2,gpt7' not found.
The second message doesn't make sense as that is the device that contains F38 that I had booted into before doing the restart, and the device I've booted into to send this email.
I did do a bios firmware update about 6 weeks ago and had to update the efi keys so that I could boot into the nvidia drivers in F38, but I would have thought that if that was causing a problem it would have surfaced before now.
When I get these errors, if I do a hard reset, by pressing the reset button on the front of the PC, then I can boot into any entry in the grub menu without issue. These errors prevent me from booting back into F38, Ubuntu and Windows.
Can anyone suggest what I need to look at to try to determine why this error is occurring?
regards, Steve
On Sun, 27 Aug 2023 10:53:48 +1000 Stephen Morris samorris@netspace.net.au wrote:
Caveat: I don't have any knowledge about the tpm and grub2 interaction.
Can anyone suggest what I need to look at to try to determine why this error is occurring?
It sounds like a bug. I think the messages are being generated because when the system tries to mount the restarted partition, it is already mounted because of the restart, and so you get both errors. The inability to mount the already mounted partition triggers the unknown error. How you would debug this is problematic, but it is a theory to consider at least. Why it would suddenly start happening is another question. What were the updates immediately before this started happening? Look in var/log/dnf.rpm.log. If you find a likely culprit, try downgrading it, or, if that doesn't work, go to koji and download an older version and downgrade it locally. https://koji.fedoraproject.org/koji/packages dnf -C downgrade [downloaded package name]
Finally, have you actually had a successful warm restart since the firmware update? That is, can you really cross the firmware update off the list of possibilities?
On 08/27/2023 10:17 AM, stan via users wrote:
It sounds like a bug. I think the messages are being generated because when the system tries to mount the restarted partition, it is already mounted because of the restart, and so you get both errors.
Shouldn't the journal show that it was mounted during the restart?
On 8/27/23 09:17, stan via users wrote:
On Sun, 27 Aug 2023 10:53:48 +1000 Stephen Morris samorris@netspace.net.au wrote:
Caveat: I don't have any knowledge about the tpm and grub2 interaction.
Can anyone suggest what I need to look at to try to determine why this error is occurring?
It sounds like a bug. I think the messages are being generated because when the system tries to mount the restarted partition, it is already mounted because of the restart, and so you get both errors. The
There's no way for the partition to be mounted. That doesn't even make sense. The OS isn't running, so "mounted" isn't a thing.
On 8/26/23 17:53, Stephen Morris wrote:
Hi, Since the last system update I have been getting TPM errors on every entry in my grub menu when I do a warm restart from F38, and this includes the menu entry for the kernel I used to boot into F38. I am getting multiple occurrences of the following message: error: ../../grub-core/commands/efi/tpm.c:150:unknown TPM error.
I am also getting the message: error: ../../grub-core/net/net.c:1552:disk 'hd2,gpt7' not found.
The second message doesn't make sense as that is the device that contains F38 that I had booted into before doing the restart, and the device I've booted into to send this email.
Those errors do suggest an issue with the UEFI BIOS.
If you open the grub console instead of trying to boot an entry, what happens when you try to look at the drive. Can you see it? What partitions does it show?
On Sun, 27 Aug 2023 17:53:24 -0700 Samuel Sieb samuel@sieb.net wrote:
On 8/27/23 09:17, stan via users wrote:
On Sun, 27 Aug 2023 10:53:48 +1000 Stephen Morris samorris@netspace.net.au wrote:
Caveat: I don't have any knowledge about the tpm and grub2 interaction.
Can anyone suggest what I need to look at to try to determine why this error is occurring?
It sounds like a bug. I think the messages are being generated because when the system tries to mount the restarted partition, it is already mounted because of the restart, and so you get both errors. The
There's no way for the partition to be mounted. That doesn't even make sense. The OS isn't running, so "mounted" isn't a thing.
Sure, what you say makes sense, but then why is the partition not being recognized? Is it somehow marked as in use because of the suspend? It is being recognized during regular boot, so what is different during a warm restart?
On Sun, 27 Aug 2023 10:59:26 -0600 Joe Zeff joe@zeff.us wrote:
On 08/27/2023 10:17 AM, stan via users wrote:
It sounds like a bug. I think the messages are being generated because when the system tries to mount the restarted partition, it is already mounted because of the restart, and so you get both errors.
Shouldn't the journal show that it was mounted during the restart?
As Samuel pointed out, this is long before any system is running, so nothing is saved in the journal because it isn't available yet when these errors occur.
On 28/8/23 10:55, Samuel Sieb wrote:
On 8/26/23 17:53, Stephen Morris wrote:
Hi, Since the last system update I have been getting TPM errors on every entry in my grub menu when I do a warm restart from F38, and this includes the menu entry for the kernel I used to boot into F38. I am getting multiple occurrences of the following message: error: ../../grub-core/commands/efi/tpm.c:150:unknown TPM error.
I am also getting the message: error: ../../grub-core/net/net.c:1552:disk 'hd2,gpt7' not found.
The second message doesn't make sense as that is the device that contains F38 that I had booted into before doing the restart, and the device I've booted into to send this email.
Those errors do suggest an issue with the UEFI BIOS.
If you open the grub console instead of trying to boot an entry, what happens when you try to look at the drive. Can you see it? What partitions does it show?
To answer Stan's question from earlier, I've had lots of warm start reboots since updating the bios and adding in the keys for the nvidia drivers. Trying to identify which package may be causing the issue might be problematic, as I was on holidays for 6 weeks and did an update when I got back, which updated around 350 packages. I'll check the grub console when I reboot my system. One other thing I forgot to mention, there is also an entry in the menu to boot into a UEFI shell, and when I try to boot into that it also gets the tpm errors.
Just as an off-topic question, hd2 is my solid state drive containing windows drive C, the UEFI partitions for Windows, F38 and Ubuntu, and the F38 and Ubuntu /boot partitions. That drive is plugged into the first physical port on the motherboard, so why does F38 not see it as hd0? The two drives it sees as hd0 and hd1 are plugged into ports 3 - 6 (I've got 4 3 TB hard drives).
regards, Steve
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On 29/8/23 08:43, Stephen Morris wrote:
On 28/8/23 10:55, Samuel Sieb wrote:
On 8/26/23 17:53, Stephen Morris wrote:
Hi, Since the last system update I have been getting TPM errors on every entry in my grub menu when I do a warm restart from F38, and this includes the menu entry for the kernel I used to boot into F38. I am getting multiple occurrences of the following message: error: ../../grub-core/commands/efi/tpm.c:150:unknown TPM error.
I am also getting the message: error: ../../grub-core/net/net.c:1552:disk 'hd2,gpt7' not found.
The second message doesn't make sense as that is the device that contains F38 that I had booted into before doing the restart, and the device I've booted into to send this email.
Those errors do suggest an issue with the UEFI BIOS.
If you open the grub console instead of trying to boot an entry, what happens when you try to look at the drive. Can you see it? What partitions does it show?
Having done a warm boot and gotten the tpm error, I opened the grub console and issued the ls subcommand which showed that what the system sees as hd2 on a normal boot, when the tpm error occurs "grub/system" is seeing that same drive as hd0. Hence it sees (hd0,gpt1) through (hd0,gp9) and for hd1 through hd4 it only sees gpt1, which matches the ssd drive and the four hard disks which only 1 partition each. What I don't understand is when the error occurs, why grub is seeing the physical drives in the order that I would expect them go be given the way they are physically plugged into the motherboard. And more importantly, what component update is causing this issue? The one thing I haven't tried yet is for a normal boot, booting off an older kernel to see if it gets the issue, and if not, the issue is potentially the current kernel?
regards, Steve
To answer Stan's question from earlier, I've had lots of warm start reboots since updating the bios and adding in the keys for the nvidia drivers. Trying to identify which package may be causing the issue might be problematic, as I was on holidays for 6 weeks and did an update when I got back, which updated around 350 packages. I'll check the grub console when I reboot my system. One other thing I forgot to mention, there is also an entry in the menu to boot into a UEFI shell, and when I try to boot into that it also gets the tpm errors.
Just as an off-topic question, hd2 is my solid state drive containing windows drive C, the UEFI partitions for Windows, F38 and Ubuntu, and the F38 and Ubuntu /boot partitions. That drive is plugged into the first physical port on the motherboard, so why does F38 not see it as hd0? The two drives it sees as hd0 and hd1 are plugged into ports 3 - 6 (I've got 4 3 TB hard drives).
regards, Steve
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On 29/8/23 08:43, Stephen Morris wrote:
On 28/8/23 10:55, Samuel Sieb wrote:
On 8/26/23 17:53, Stephen Morris wrote:
Hi, Since the last system update I have been getting TPM errors on every entry in my grub menu when I do a warm restart from F38, and this includes the menu entry for the kernel I used to boot into F38. I am getting multiple occurrences of the following message: error: ../../grub-core/commands/efi/tpm.c:150:unknown TPM error.
I am also getting the message: error: ../../grub-core/net/net.c:1552:disk 'hd2,gpt7' not found.
The second message doesn't make sense as that is the device that contains F38 that I had booted into before doing the restart, and the device I've booted into to send this email.
Those errors do suggest an issue with the UEFI BIOS.
If you open the grub console instead of trying to boot an entry, what happens when you try to look at the drive. Can you see it? What partitions does it show?
If when the tpm error occurs, I edit the grub menu entry and change all occurrences of hd2 to hd0, even though the tpm error still occur, I can boot into any one of the kernel entries, but a subsequent warm boot still gets the tpm errors. Hence the issue appears to not be kernel related, but may be a grub issue. I'll try rebuilding the grub menus to see if that makes any difference.
regards, Steve
To answer Stan's question from earlier, I've had lots of warm start reboots since updating the bios and adding in the keys for the nvidia drivers. Trying to identify which package may be causing the issue might be problematic, as I was on holidays for 6 weeks and did an update when I got back, which updated around 350 packages. I'll check the grub console when I reboot my system. One other thing I forgot to mention, there is also an entry in the menu to boot into a UEFI shell, and when I try to boot into that it also gets the tpm errors.
Just as an off-topic question, hd2 is my solid state drive containing windows drive C, the UEFI partitions for Windows, F38 and Ubuntu, and the F38 and Ubuntu /boot partitions. That drive is plugged into the first physical port on the motherboard, so why does F38 not see it as hd0? The two drives it sees as hd0 and hd1 are plugged into ports 3 - 6 (I've got 4 3 TB hard drives).
regards, Steve
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Tue, 29 Aug 2023 18:08:59 +1000 Stephen Morris samorris@netspace.net.au wrote:
Having done a warm boot and gotten the tpm error, I opened the grub console and issued the ls subcommand which showed that what the system sees as hd2 on a normal boot, when the tpm error occurs "grub/system" is seeing that same drive as hd0. Hence it sees (hd0,gpt1) through (hd0,gp9) and for hd1 through hd4 it only sees gpt1, which matches the ssd drive and the four hard disks which only 1 partition each. What I don't understand is when the error occurs, why grub is seeing
^ not?
the physical drives in the order that I would expect them go be given the way they are physically plugged into the motherboard. And more importantly, what component update is causing this issue?
Was grub updated?
The one thing I haven't tried yet is for a normal boot, booting off an older kernel to see if it gets the issue, and if not, the issue is potentially the current kernel?
Yes that would be a good test. There have been a lot of changes to the fedora kernel SPEC file to clean it up, and streamline it. It isn't impossible that you are seeing a corner case side effect of that, though unlikely.
regards, Steve
To answer Stan's question from earlier, I've had lots of warm start reboots since updating the bios and adding in the keys for the nvidia drivers. Trying to identify which package may be causing the issue might be problematic, as I was on holidays for 6 weeks and did an update when I got back, which updated around 350 packages. I'll check the grub console when I reboot my system. One other thing I forgot to mention, there is also an entry in the menu to boot into a UEFI shell, and when I try to boot into that it also gets the tpm errors.
Just as an off-topic question, hd2 is my solid state drive containing windows drive C, the UEFI partitions for Windows, F38 and Ubuntu, and the F38 and Ubuntu /boot partitions. That drive is plugged into the first physical port on the motherboard, so why does F38 not see it as hd0? The two drives it sees as hd0 and hd1 are plugged into ports 3 - 6 (I've got 4 3 TB hard drives).
I don't have an answer, but I wonder if there is an obscure setting in the bios that is responsible. What is the boot order set to?
On 30/8/23 01:54, stan via users wrote:
On Tue, 29 Aug 2023 18:08:59 +1000 Stephen Morris samorris@netspace.net.au wrote:
Having done a warm boot and gotten the tpm error, I opened the grub console and issued the ls subcommand which showed that what the system sees as hd2 on a normal boot, when the tpm error occurs "grub/system" is seeing that same drive as hd0. Hence it sees (hd0,gpt1) through (hd0,gp9) and for hd1 through hd4 it only sees gpt1, which matches the ssd drive and the four hard disks which only 1 partition each. What I don't understand is when the error occurs, why grub is seeing
^ not?the physical drives in the order that I would expect them go be given the way they are physically plugged into the motherboard. And more importantly, what component update is causing this issue?
Was grub updated?
The one thing I haven't tried yet is for a normal boot, booting off an older kernel to see if it gets the issue, and if not, the issue is potentially the current kernel?
Yes that would be a good test. There have been a lot of changes to the fedora kernel SPEC file to clean it up, and streamline it. It isn't impossible that you are seeing a corner case side effect of that, though unlikely.
regards, Steve
To answer Stan's question from earlier, I've had lots of warm start reboots since updating the bios and adding in the keys for the nvidia drivers. Trying to identify which package may be causing the issue might be problematic, as I was on holidays for 6 weeks and did an update when I got back, which updated around 350 packages. I'll check the grub console when I reboot my system. One other thing I forgot to mention, there is also an entry in the menu to boot into a UEFI shell, and when I try to boot into that it also gets the tpm errors.
Just as an off-topic question, hd2 is my solid state drive containing windows drive C, the UEFI partitions for Windows, F38 and Ubuntu, and the F38 and Ubuntu /boot partitions. That drive is plugged into the first physical port on the motherboard, so why does F38 not see it as hd0? The two drives it sees as hd0 and hd1 are plugged into ports 3 - 6 (I've got 4 3 TB hard drives).
I don't have an answer, but I wonder if there is an obscure setting in the bios that is responsible. What is the boot order set to?
The bios is set to boot off my ssd drive, which is the first drive plugged into the motherboard, which is the device that Fedora sees as hd2.
I did a system update yesterday, which upgraded the kernel to 6.4.12 and also updated grub, and then updated the grub menus via grub2-mkconfig as I always do, and that has not made any difference to the issue. I have grub configured to build sub-menus for all the kernel entries as well as showing the latest kernel in the main menu, so that I have all the Fedora kernels and Ubuntu kernels in sub-menus. What I have now found is that if I open up a sub-menu, that is when the tpm error occurs, and since the grub update it is now producing an extra error telling me to load a kernel first (what I don't understand is that message seems to be coming from an I386 sub-folder but my environment is 64 bit, or does that mean that somehow or other grub has reverted to 32 bit?). I've also mentioned in another thread, that if when I get the tpm errors I edit the grub menu entry and change all occurrences of hd2 to hd0, even though it continues to display the tpm errors it successfully boots into F38. It seems as though at the moment it boots normally if I select a main menu entry to boot from, but only if the tpm error hasn't already occurred. If the tpm error has occurred none of the menu entries will boot, which includes the Chainloader entry for Windows.
Having started my machine from a cold start, when the grub menu's were displayed, I went to the grub command line and issued the LS command to list all devices, that showed my boot device as hd0 (hd0,gpt1 - hd0,gpt9), and then when I exited from the command line, and selected the menu entry for the latest Fedora kernel, which specified to boot from hd2,gpt7 (this is the fedora UEFI partition), it successfully booted into Fedora. How is this possible when the grub command line is indicating that grub is seeing the devices differently? What I might add to this is that the way the grub command line is showing the devices is the way I would expect them to be shown given the way the devices are physically connected to the motherboard.
regards, Steve
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Wed, 30 Aug 2023 09:32:22 +1000 Stephen Morris samorris@netspace.net.au wrote:
The bios is set to boot off my ssd drive, which is the first drive plugged into the motherboard, which is the device that Fedora sees as hd2.
I did a system update yesterday, which upgraded the kernel to 6.4.12 and also updated grub, and then updated the grub menus via grub2-mkconfig as I always do, and that has not made any difference to the issue. I have grub configured to build sub-menus for all the kernel entries as well as showing the latest kernel in the main menu, so that I have all the Fedora kernels and Ubuntu kernels in sub-menus. What I have now found is that if I open up a sub-menu, that is when the tpm error occurs, and since the grub update it is now producing an extra error telling me to load a kernel first (what I don't understand is that message seems to be coming from an I386 sub-folder but my environment is 64 bit, or does that mean that somehow or other grub has reverted to 32 bit?). I've also mentioned in another thread, that if when I get the tpm errors I edit the grub menu entry and change all occurrences of hd2 to hd0, even though it continues to display the tpm errors it successfully boots into F38. It seems as though at the moment it boots normally if I select a main menu entry to boot from, but only if the tpm error hasn't already occurred. If the tpm error has occurred none of the menu entries will boot, which includes the Chainloader entry for Windows.
Having started my machine from a cold start, when the grub menu's were displayed, I went to the grub command line and issued the LS command to list all devices, that showed my boot device as hd0 (hd0,gpt1 - hd0,gpt9), and then when I exited from the command line, and selected the menu entry for the latest Fedora kernel, which specified to boot from hd2,gpt7 (this is the fedora UEFI partition), it successfully booted into Fedora. How is this possible when the grub command line is indicating that grub is seeing the devices differently? What I might add to this is that the way the grub command line is showing the devices is the way I would expect them to be shown given the way the devices are physically connected to the motherboard.
I understand what you are asking, and it is certainly a conundrum, but I have no insight to offer. Maybe open a bugzilla against grub2. I don't think it is the problem, but the people who maintain grub2 probably have a good understanding of this part of the boot process, and might be able to point to the real culprit.
On 30/8/23 23:12, stan via users wrote:
On Wed, 30 Aug 2023 09:32:22 +1000 Stephen Morris samorris@netspace.net.au wrote:
The bios is set to boot off my ssd drive, which is the first drive plugged into the motherboard, which is the device that Fedora sees as hd2.
I did a system update yesterday, which upgraded the kernel to 6.4.12 and also updated grub, and then updated the grub menus via grub2-mkconfig as I always do, and that has not made any difference to the issue. I have grub configured to build sub-menus for all the kernel entries as well as showing the latest kernel in the main menu, so that I have all the Fedora kernels and Ubuntu kernels in sub-menus. What I have now found is that if I open up a sub-menu, that is when the tpm error occurs, and since the grub update it is now producing an extra error telling me to load a kernel first (what I don't understand is that message seems to be coming from an I386 sub-folder but my environment is 64 bit, or does that mean that somehow or other grub has reverted to 32 bit?). I've also mentioned in another thread, that if when I get the tpm errors I edit the grub menu entry and change all occurrences of hd2 to hd0, even though it continues to display the tpm errors it successfully boots into F38. It seems as though at the moment it boots normally if I select a main menu entry to boot from, but only if the tpm error hasn't already occurred. If the tpm error has occurred none of the menu entries will boot, which includes the Chainloader entry for Windows.
Having started my machine from a cold start, when the grub menu's were displayed, I went to the grub command line and issued the LS command to list all devices, that showed my boot device as hd0 (hd0,gpt1 - hd0,gpt9), and then when I exited from the command line, and selected the menu entry for the latest Fedora kernel, which specified to boot from hd2,gpt7 (this is the fedora UEFI partition), it successfully booted into Fedora. How is this possible when the grub command line is indicating that grub is seeing the devices differently? What I might add to this is that the way the grub command line is showing the devices is the way I would expect them to be shown given the way the devices are physically connected to the motherboard.
I understand what you are asking, and it is certainly a conundrum, but I have no insight to offer. Maybe open a bugzilla against grub2. I don't think it is the problem, but the people who maintain grub2 probably have a good understanding of this part of the boot process, and might be able to point to the real culprit.
Thanks Stan, I'll do that.
regards, Steve
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
-
Joe Zeff -
Samuel Sieb -
stan -
Stephen Morris