Skip to contentSkip to conversation

v5.9-rc1 commit reliably breaks pci nvme detection

DISCUSSION Ahmed S. Darwish 6 participants 9 messages Aug 17, 2020 – Aug 21, 2020 NVMe

lore
Share & tools
lore
#0 lore raw

Hello,

Below v5.9-rc1 commit reliably breaks my boot on a Thinkpad e480 laptop. PCI nvme detection fails, and the kernel becomes not able anymore to find the rootfs / parse "root=".

Bisecting v5.8=>v5.9-rc1 blames that commit. Reverting it *reliably* fixes the problem and makes me able to boot v5.9-rc1.

Please advise.

commit 61f3b89630973037f67d8e25e5d26e80a51a7b37

Author: Chaitanya Kulkarni <[email protected]>
Date:   Wed Jun 17 10:05:13 2020 +0200

nvme-pci: use unsigned for io queue depth

The NVMe PCIe declares module parameter io_queue_depth as int. Change this to u16 as queue depth can never be negative. Now to reflect this update module parameter getter function from param_get_int() -> param_get_uint() and respective setter function with type of n changed from int to u16 with param_set_int() to param_set_ushort(). Finally update struct nvme_dev q_depth member to u16 and use u16 in min_t() when calculating dev->q_depth in the nvme_pci_enable() (since q_depth is now u16) and use unsigned int instead of int when calculating dev->tagset.queue_depth as target variable tagset->queue_depth is of type unsigned int in nvme_dev_add().

Thanks,

--
Ahmed S. Darwish
Linutronix GmbH

maintainer
#1 lore raw
Quoted 8 lines · Ahmed S. Darwish

Hello,

Below v5.9-rc1 commit reliably breaks my boot on a Thinkpad e480 laptop. PCI nvme detection fails, and the kernel becomes not able anymore to find the rootfs / parse "root=".

Bisecting v5.8=>v5.9-rc1 blames that commit. Reverting it *reliably* fixes the problem and makes me able to boot v5.9-rc1.

The fix is staged in the nvme tree here:

http://git.infradead.org/nvme.git/commit/286155561ecd13b6c85a78eaf2880d3baea03b9e

_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme

#2 lore raw
Quoted 7 lines · Keith Busch
Quoted 6 lines · Ahmed S. Darwish

Hello,

Below v5.9-rc1 commit reliably breaks my boot on a Thinkpad e480 laptop. PCI nvme detection fails, and the kernel becomes not able anymore to find the rootfs / parse "root=".

Quoted 6 lines
Quoted 2 lines

Bisecting v5.8=>v5.9-rc1 blames that commit. Reverting it *reliably* fixes the problem and makes me able to boot v5.9-rc1.

The fix is staged in the nvme tree here:

http://git.infradead.org/nvme.git/commit/286155561ecd13b6c85a78eaf2880d3baea03b9e

That would have been nice to have in -rc1...

Signature · 7 lines
Jens Axboe


_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
maintainer
#3 lore raw
Quoted 8 lines · Jens Axboe
Quoted 7 lines · Keith Busch
Quoted 6 lines · Ahmed S. Darwish

Hello,

Below v5.9-rc1 commit reliably breaks my boot on a Thinkpad e480 laptop. PCI nvme detection fails, and the kernel becomes not able anymore to find the rootfs / parse "root=".

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

Bisecting v5.8=>v5.9-rc1 blames that commit. Reverting it *reliably* fixes the problem and makes me able to boot v5.9-rc1.

The fix is staged in the nvme tree here:

http://git.infradead.org/nvme.git/commit/286155561ecd13b6c85a78eaf2880d3baea03b9e

That would have been nice to have in -rc1...

And now we're getting very close to shipping items for -rc2, and it's still not in. Can we please get the nvme pull request out for -rc2?

Signature · 1 line
Jens Axboe
#4 lore raw
Quoted 12 lines · Jens Axboe
Quoted 11 lines · Jens Axboe
Quoted 10 lines · Keith Busch
Quoted 8 lines · Ahmed S. Darwish

Hello,

Below v5.9-rc1 commit reliably breaks my boot on a Thinkpad e480 laptop. PCI nvme detection fails, and the kernel becomes not able anymore to find the rootfs / parse "root=".

Bisecting v5.8=>v5.9-rc1 blames that commit. Reverting it *reliably* fixes the problem and makes me able to boot v5.9-rc1.

Quoted 9 lines
Quoted 5 lines
Quoted 3 lines

The fix is staged in the nvme tree here:

http://git.infradead.org/nvme.git/commit/286155561ecd13b6c85a78eaf2880d3baea03b9e

That would have been nice to have in -rc1...

And now we're getting very close to shipping items for -rc2, and it's still not in. Can we please get the nvme pull request out for -rc2?

I keep wondering that myself. Completely breaking the boot like this is really not nice -- and for x86-64 laptops no less :-(

The fix is really small and isolated. "Urgent pull requests", containing only a fix or two, were created *exactly* for this reson...

Thanks,

--
Ahmed S. Darwish
Linutronix GmbH

_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme

maintainer
#5 lore raw
Quoted 17 lines · Ahmed S. Darwish
Quoted 16 lines · Jens Axboe
Quoted 15 lines · Jens Axboe
Quoted 13 lines · Keith Busch
Quoted 8 lines · Ahmed S. Darwish

Hello,

Below v5.9-rc1 commit reliably breaks my boot on a Thinkpad e480 laptop. PCI nvme detection fails, and the kernel becomes not able anymore to find the rootfs / parse "root=".

Bisecting v5.8=>v5.9-rc1 blames that commit. Reverting it *reliably* fixes the problem and makes me able to boot v5.9-rc1.

The fix is staged in the nvme tree here:

http://git.infradead.org/nvme.git/commit/286155561ecd13b6c85a78eaf2880d3baea03b9e

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

That would have been nice to have in -rc1...

And now we're getting very close to shipping items for -rc2, and it's still not in. Can we please get the nvme pull request out for -rc2?

I keep wondering that myself. Completely breaking the boot like this is really not nice -- and for x86-64 laptops no less :-(

To be fair, I've only heard this one complaint about it, so hopefully it's not too widespread. I'm on an x86-64 laptop myself with nvme, and it works just fine :-)

Quoted 2 lines

The fix is really small and isolated. "Urgent pull requests", containing only a fix or two, were created *exactly* for this reson...

Totally agree on that, it should have gone in for -rc1. This will be going upstream tomorrow, so the end is near...

Signature · 1 line
Jens Axboe
#6 lore raw
Quoted 3 lines · Jens Axboe

To be fair, I've only heard this one complaint about it, so hopefully it's not too widespread. I'm on an x86-64 laptop myself with nvme, and it works just fine :-)

The cause for this is the weird NVMe of by ones, where 0 in a field means 1. So for the overflow to happen you need a controller that supports USHORT_MAX queue entries. Which don't seem to be all that common.

_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme

maintainer
#7 lore raw
Quoted 9 lines · Christoph Hellwig
Quoted 3 lines · Jens Axboe

To be fair, I've only heard this one complaint about it, so hopefully it's not too widespread. I'm on an x86-64 laptop myself with nvme, and it works just fine :-)

The cause for this is the weird NVMe of by ones, where 0 in a field means 1. So for the overflow to happen you need a controller that supports USHORT_MAX queue entries. Which don't seem to be all that common.

Yeah, don't think I've ever seen those. I come across 1023 and 128 all the time, but I don't have one in my arsenal of NVMe drives that is any different than those two.

Signature · 7 lines
Jens Axboe


_______________________________________________
Linux-nvme mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-nvme
#8 lore raw
In reply to Christoph Hellwig
Quoted 6 lines · Christoph Hellwig
Quoted 3 lines · Jens Axboe

To be fair, I've only heard this one complaint about it, so hopefully it's not too widespread. I'm on an x86-64 laptop myself with nvme, and it works just fine:-)

The cause for this is the weird NVMe of by ones, where 0 in a field means 1.

I thought that this was a common trick by spec writers to fit a number in range (0, 2^x] in x bits (as opposed to x+1).