Thanks for the bug report.

I see your report on MLO forum (https://www.mageialinux-online.org/forum/topic-27967+probleme-install-mageia-7-sur-ssd-m2.php)

As I see, you tried to format your SSD drive looking for bad block with M7.1 Classic x86_64 ISO. New hardware usually can't be properly supported after ISO release.

Best advises had been addressed:
- Be sure to install in UEFI Mode (Disabling Legacy BIOS/CSM Support and Secure Boot as we don't support the latter)
- try M8 Beta1 install => OK
- try a Live M7 ISO (or even Gparted live ISO from https://gparted.org/liveusb.php) and use GParted to format before reinstall with M7.1 ISO Classic without formatting and without looking for Bad Blocks.

Meanwhile, assigning to Mageia Tools Maintainers.
Packagers: Please change status to "Assigned" if you work on it, feel free to close this as Cauldron seems to be not affected.

@Reporter, feel free to add further details by adding attachments.

Assignee: bugsquad => mageiatools
CC: (none) => ouaurelien
Keywords: (none) => Triaged
Status comment: (none) => OP seems to have run M8 Cauldron successfully on his system.
Source RPM: (none) => DrakX Installer

This is almost certainly a kernel/driver issue. The kernel on the Mageia 7.1 classical installer and Live ISOs is too old to support your new hardware. Please try using the network installer ISO instead, which was updated in August to use a more recent kernel.

CC: (none) => mageia

Yeah, Martin is right.
OP reported: 5.1.14-desktop-1.mga7 for Kernel version.

@reporter, you can try netinstall ISO from this mirror in France:
http://ftp.free.fr/mirrors/mageia.org/distrib/7.1/x86_64/install/images/

Created attachment 11912 [details]
full bug report on this :)

well, I've tried what was suggested, installing mageia 7.1 from a netinstall cdrom.

all goes well until I've partitioned the SSD M2 for a /boot and / and wanted to format the parts in ext4. 

I got the bug and this time was able to get the full bug report using the 'bug' command, so here it is ^^)

CC: (none) => jfbachelet

Created attachment 11913 [details]
same bug with mageia 8b1 install dvd :(

bad news but the bug here on mageia 8b1 install too :(

See if the suggestions in
https://bbs.archlinux.org/viewtopic.php?id=249098
particularly from comment 4, adding the kernel parameter
nvme_core.default_ps_max_latency_us=5500

CC: (none) => davidwhodgins

Sorry, hadn't read that bug report fully before posting the above. According
to that link, it's defective hardware.

If Windows is installed and working on that drive, it seems unlikely the hardware is defective. From the log

<4>[  597.464390] nvme nvme0: I/O 384 QID 5 timeout, aborting

so it's worth trying adding the nvme_core.default_ps_max_latency_us=5500 kernel parameter on the boot command line.

Other than that, this is a kernel/driver issue, not an installer bug.

Status comment: OP seems to have run M8 Cauldron successfully on his system. => (none)
Whiteboard: (none) => MGA7TOO
Component: Installer => RPM Packages
Source RPM: DrakX Installer => kernel-5.7.14-desktop-1.mga7, kernel-5.7.9-desktop-1.mga8
Assignee: mageiatools => kernel
Version: 7 => Cauldron

Created attachment 11914 [details]
report after adding boot stuf

Well, I've tried what you said, adding the 'nvme_core.default_ps_max_latency_us=5500' line to the boot command line, and that changed nothing :(

BTW, I've waited way longer that time and got a fuller bug report.

and, Yes, windows is installed on that same ssd and it's working well, I had no problems to format its windows partitions to NTFS.

there are no problems too if I use the F2FS filesystem on that mageia / root partition on the SSD, it format quickly and installation goes till the end.

Note this is a hardware-specific bug. I've installed from the 7.1 classical installer onto an AMD Ryzen 2400G + ASUS X470i + Samsung M.2 SSD 970 EVO Plus system, using the ext4 filesystem, with no problems.

just to be precise : the SSD M2 is to be shared by windows 10 pro (that is already installed on the first half of it, and mageia on the second part that I have problems to install due to this bug. :(

the computer configuration is this : rizen 9, msi x570-A-Pro mobo, ssd M2  Integral INSSD1TM280NM1 1 TO in the bottom M2 port of mobo (the top one is for a lightning SSD, currently empty)

Could you try default_ps_max_latency_us=0 to entirely disable APST?

CC: (none) => pterjan

Created attachment 11918 [details]
results with disabled APST

Changed the boot line again for disabling APST like suggested, bug occur too but don't hang very long the machine.

forget to put that too :

'Error: formating of /dev/nvme0n1p7 failed at /usr/lib/libDraX/fs/format-pm line 285'

found this on ctrl-alt-F1 sehll

Whatever the trigger is, there is a kernel bug 30s before the problem (which was already in your previous logs but I missed):

<4>[  131.687109] refcount_t: underflow; use-after-free.
<4>[  131.687125] WARNING: CPU: 19 PID: 0 at lib/refcount.c:28 refcount_warn_saturate+0xa6/0xf0
<4>[  131.687125] Modules linked in: uas usb_storage dm_zero dm_mirror dm_region_hash dm_log dm_mod af_packet psmouse squashfs loop nls_utf8 isofs ide_cd_mod ide_core evdev hid_generic usbhid hid wmi_bmof xhci_pci xhci_hcd crc32_pclmul crc32c_intel aesni_intel snd_pcm usbcore glue_helper crypto_simd cryptd r8169 snd_timer snd realtek libphy i2c_piix4 soundcore usb_common sr_mod wmi acpi_cpufreq
<4>[  131.687145] CPU: 19 PID: 0 Comm: swapper/19 Not tainted 5.7.9-desktop-1.mga8 #1
<4>[  131.687146] Hardware name: Micro-Star International Co., Ltd. MS-7C37/X570-A PRO (MS-7C37), BIOS H.70 01/09/2020
<4>[  131.687148] RIP: 0010:refcount_warn_saturate+0xa6/0xf0
<4>[  131.687150] Code: 05 cd 90 fe 00 01 e8 20 c8 b2 ff 0f 0b c3 80 3d bb 90 fe 00 00 75 95 48 c7 c7 c0 44 13 a9 c6 05 ab 90 fe 00 01 e8 01 c8 b2 ff <0f> 0b c3 80 3d 9a 90 fe 00 00 0f 85 72 ff ff ff 48 c7 c7 18 45 13
<4>[  131.687151] RSP: 0018:ffffa0e080600ee0 EFLAGS: 00010082
<4>[  131.687152] RAX: 0000000000000000 RBX: ffff90a3b2985858 RCX: 0000000000000007
<4>[  131.687153] RDX: 00000000fffffff8 RSI: 0000000000000096 RDI: ffff90a3becd98e0
<4>[  131.687154] RBP: ffff90a3ae74bd00 R08: 0000000000000001 R09: 00000000000003f3
<4>[  131.687154] R10: 00000000000153ec R11: 0000000000000001 R12: ffffc0e07efc5200
<4>[  131.687155] R13: 0000000000000013 R14: ffffa0e080600f9c R15: ffff90a3ae74bd00
<4>[  131.687157] FS:  0000000000000000(0000) GS:ffff90a3becc0000(0000) knlGS:0000000000000000
<4>[  131.687158] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
<4>[  131.687158] CR2: 000000000460cdf0 CR3: 000000009440a000 CR4: 0000000000340ee0
<4>[  131.687159] Call Trace:
<4>[  131.687161]  <IRQ>
<4>[  131.687164]  blk_mq_complete_request+0x6c/0x100
<4>[  131.687168]  nvme_process_cq+0xee/0x200
<4>[  131.687170]  nvme_irq+0x10/0x20
<4>[  131.687173]  __handle_irq_event_percpu+0x45/0x190
<4>[  131.687175]  handle_irq_event+0x4c/0xa7
<4>[  131.687178]  handle_edge_irq+0x8b/0x240
<4>[  131.687180]  do_IRQ+0x50/0xe0
<4>[  131.687182]  common_interrupt+0xf/0xf
<4>[  131.687183]  </IRQ>

Also, the image you used is using a quite old kernel from July (5.7.9-desktop-1.mga8). Before reporting this upstream, could you try with one from
http://ftp.free.fr/mirrors/mageia.org/distrib/cauldron/x86_64/install/images/ which use 5.8.14-desktop-1.mga8?

Thanks

Created attachment 11919 [details]
a try with gparted from live dvd

ok Pascal I try this latest one.

BTW, I've just made a try with M8b1 Live DVD (from July 2020, 19)
here is the gparted report, it failed too,

what I've done is booting the live, and updated it fully, then tried with Drakdisk, it failed :( then I wanted to try gparted but it was not seeing the nvme disk anymore :(

so rebooted and done the update again and tried again gparted, it failed too, see its report.

Created attachment 11920 [details]
try with latest kernels images netinstall

here is the result of the try with latest image as you've suggested, It failed again :(

Thanks for trying!

Latest trace for easy access:

<4>[  145.998576] ------------[ cut here ]------------
<4>[  145.998577] refcount_t: underflow; use-after-free.
<4>[  145.998591] WARNING: CPU: 17 PID: 0 at lib/refcount.c:28 refcount_warn_saturate+0xa6/0xf0
<4>[  145.998592] Modules linked in: dm_zero dm_mirror dm_region_hash dm_log dm_mod ide_cd_mod ide_core psmouse ipv6 crc_ccitt squashfs loop af_packet evdev hid_generic usbhid hid wmi_bmof xhci_pci xhci_hcd crc32_pclmul crc32c_intel snd_pcm aesni_intel usbcore glue_helper crypto_simd cryptd snd_timer r8169 snd i2c_piix4 realtek libphy soundcore sr_mod usb_common wmi acpi_cpufreq
<4>[  145.998608] CPU: 17 PID: 0 Comm: swapper/17 Not tainted 5.8.14-desktop-1.mga8 #1
<4>[  145.998609] Hardware name: Micro-Star International Co., Ltd. MS-7C37/X570-A PRO (MS-7C37), BIOS H.70 01/09/2020
<4>[  145.998611] RIP: 0010:refcount_warn_saturate+0xa6/0xf0
<4>[  145.998612] Code: 05 83 d5 fd 00 01 e8 6f 13 b2 ff 0f 0b c3 80 3d 71 d5 fd 00 00 75 95 48 c7 c7 20 d9 13 9b c6 05 61 d5 fd 00 01 e8 50 13 b2 ff <0f> 0b c3 80 3d 50 d5 fd 00 00 0f 85 72 ff ff ff 48 c7 c7 78 d9 13
<4>[  145.998613] RSP: 0018:ffffa15780598ef8 EFLAGS: 00010082
<4>[  145.998615] RAX: 0000000000000000 RBX: ffff94196ba8f800 RCX: ffff94197ec58928
<4>[  145.998615] RDX: 00000000ffffffd8 RSI: 0000000000000027 RDI: ffff94197ec58920
<4>[  145.998616] RBP: ffff9419676ede00 R08: 0000000000000001 R09: 00000000000003d6
<4>[  145.998617] R10: 0000000000014988 R11: 0000000000000001 R12: ffffc1577f244800
<4>[  145.998617] R13: 0000000000000011 R14: ffffa15780598fbc R15: ffff9419676ede00
<4>[  145.998619] FS:  0000000000000000(0000) GS:ffff94197ec40000(0000) knlGS:0000000000000000
<4>[  145.998619] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
<4>[  145.998620] CR2: 000000000153e5b8 CR3: 00000007c79fc000 CR4: 0000000000340ee0
<4>[  145.998621] Call Trace:
<4>[  145.998622]  <IRQ>
<4>[  145.998625]  blk_mq_force_complete_rq+0x68/0xe0
<4>[  145.998628]  blk_mq_complete_request+0xa/0x10
<4>[  145.998630]  nvme_process_cq+0xdb/0x1f0
<4>[  145.998632]  nvme_irq+0x10/0x20
<4>[  145.998635]  __handle_irq_event_percpu+0x45/0x190
<4>[  145.998637]  handle_irq_event+0x4c/0xa7
<4>[  145.998638]  handle_edge_irq+0x8b/0x240
<4>[  145.998641]  asm_call_irq_on_stack+0x12/0x20
<4>[  145.998642]  </IRQ>
<4>[  145.998644]  common_interrupt+0xb2/0x140
<4>[  145.998646]  asm_common_interrupt+0x1e/0x40
<4>[  145.998649] RIP: 0010:cpuidle_enter_state+0xb6/0x3f0
<4>[  145.998650] Code: 60 2e 7d 65 e8 bb 38 88 ff 49 89 c7 0f 1f 44 00 00 31 ff e8 6c 4f 88 ff 80 7c 24 0f 00 0f 85 d4 01 00 00 fb 66 0f 1f 44 00 00 <45> 85 e4 0f 88 e0 01 00 00 49 63 d4 4c 2b 7c 24 10 48 8d 04 52 48
<4>[  145.998651] RSP: 0018:ffffa157801afe78 EFLAGS: 00000246
<4>[  145.998652] RAX: ffff94197ec6be40 RBX: ffff94196be14800 RCX: 000000000000001f
<4>[  145.998653] RDX: 0000000000000000 RSI: 0000000021af2900 RDI: 0000000000000000
<4>[  145.998653] RBP: ffffffff9b4e3be0 R08: 00000021fe3175d8 R09: 0000000000000007
<4>[  145.998654] R10: 000000000000004b R11: 0000000000000042 R12: 0000000000000001
<4>[  145.998654] R13: ffff94196be14800 R14: 0000000000000001 R15: 00000021fe3175d8
<4>[  145.998657]  ? cpuidle_enter_state+0xa4/0x3f0
<4>[  145.998659]  cpuidle_enter+0x29/0x40
<4>[  145.998661]  do_idle+0x1f9/0x2e0
<4>[  145.998663]  cpu_startup_entry+0x19/0x20
<4>[  145.998665]  start_secondary+0x151/0x190
<4>[  145.998667]  secondary_startup_64+0xa4/0xb0
<4>[  145.998669] ---[ end trace d021a7959d46c7c9 ]---

(In reply to Pascal Terjan from comment #19)
> Thanks for trying!

You're welcome ^^)

what now ?

some more tries : 

no problems to format parts in ReiserFS or XFS on the NVME ssd.

BUT it impossible to format a 'BIOS boot or empty partition' on the NVME ssd...

oh, and if I make the /Boot and / in XFS on the NVME ssd, when it comes to finish the partitionnement Drakdisk cries for a /Bios Boot on the GPT hard disk to continue the installation...

It hasn't asked for this when the NVME ssd parts was formated as F2FS.

Are you sure you booted the installer in EFI mode? A BIOS boot partition is only needed for legacy boot. And it can't be formatted because it has no associated file system - it's just an empty space where GRUB2 can install itself.

booted the installer in efi mode yes.

when it cried it says that was 'in case you boot the machine in legacy mode'...

Please open a separate bug report for that and attach the report.bug generated by the installer 'bug' command.

restarted the install again, killed the bios boot part and continued the install 
 and this time it hasn't complained... perhaps it was really legacy boot the first time?

I think we can close this resolved invalid?

CC: (none) => fri

the problem with /boot etc, was another problem that as surfaced  and had been cured but the Ext4 formating problem of that particular SSD is still there.


as I don't have this machine anymore I can't try anything new :(

BTW, it was still impossible to format the / partition on the ssd as Ext4 without the machine hanging :(

So to deliver it to my customer I've 'work-arounded' the problem by formating the / partition on this ssd as ReiserFS, without any crash, and installed Mageia on it with the other partitions (/Home, and /user/local) on the hard disk.

configured like this it works well.

hope we can find someday why format hang when trying EXT4 on these SSDs...
as soon as I have again that kind of computer config I'll make more tests.

OK
Interesting bug
Thank you

Based on comment 29 that this machine no longer exists, and Mageia can properly handle SSDs since long time...

Closing as INVALID.

Status: NEW => RESOLVED
Resolution: (none) => INVALID

not resolved :-( the problerm is still there fall 2022.

see https://bugs.mageia.org/show_bug.cgi?id=31217

Resolution: INVALID => (none)
Status: RESOLVED => REOPENED

*** Bug 31217 has been marked as a duplicate of this bug. ***

Comments from duplicate bug:
Not reproducible due to lack of informations:

- is this disk preformatted with some partitions or is it a really empty disk?
- which steps do you chose (as already asked and not answered): https://doc.mageia.org/installer/8/en/content/diskPartitioning.html
-> Custom Disk Partitioning ->Auto Allocate->fiddling with size afterwards?
-> Use existing partitions ->fiddling with size afterwards?
-> Custom Disk Partitioning -> real custom partitioning?
-> Use Free Space on a Windows Partition->fiddling with size afterwards?
-> ???

In bug 27364 you mention that you have Windows on the same disk. Did you shut down Windows properly and made sure that your disk state is consistent and clean?

Which iso do you use? Netinstall or which other ISO?

Is your BIOS still heavily outdated as in bug 27364?

....

Comments from duplicate bug:
(In reply to Jean-François Bachelet from comment #5)
 I just create partitions the
> size I need plus one or two in sus of the usual /, /usr/local, /home and
> /swap, usualy a /vms for all the virtual machines I create and a /save for
> stocking some stuff. 
> ....
> you can try this too : let drakdisk auto partition, then just resize the
> /home part and click on next, it ask which partition should be formated,


Some important information is missing here. The partition scheme you describe in the first sentence is no "usual" scheme. This not possible to be created by any "autopartition".

And please clarify what you are doing as asked in my last comment as there is no "auto partition" mode (at least not called this way in drakdisk)
https://doc.mageia.org/installer/8/en/content/diskPartitioning.html
- Use Existing Partitions
- Use Free Space
- Use Free Space on a Windows Partition
- Custom Disk Partitioning -> Auto allocate

(In reply to sturmvogel from comment #34)
> Comments from duplicate bug:
> Not reproducible due to lack of informations:
> 
> - is this disk preformatted with some partitions or is it a really empty
> disk?

the M2 NVME disk was brand new at first as I just made that machine from all new components. and this time there is no Windows installed first on it like it was with the first 2020 machine.

> - which steps do you chose (as already asked and not answered):
> https://doc.mageia.org/installer/8/en/content/diskPartitioning.html
> -> Custom Disk Partitioning ->Auto Allocate->fiddling with size afterwards?
> -> Use existing partitions ->fiddling with size afterwards?
> -> Custom Disk Partitioning -> real custom partitioning?

real custom partitioning using drakdisk, start with empty new disk and create all parts needed by Me and Mageia. no fiddling afterward.

> -> Use Free Space on a Windows Partition->fiddling with size afterwards?
> -> ???
> 
> In bug 27364 you mention that you have Windows on the same disk. Did you
> shut down Windows properly and made sure that your disk state is consistent
> and clean?

Windoh! was installed on the first half of the M2 NVME and it never touched the second part of it. So yes the second part for Mageia was still pristine new and clean.

I always shut down windoh! properly as it is way too prone to behaving havoc if not. and it was there a brand new install of windows 10 professional for my customer on first half of the M2 NVME.

> Which iso do you use? Netinstall or which other ISO?

in 2020 full 7.1 install disk at first, then tried after with 7.1 net install, then with 8b1 net install disk, all failed.

today I use mainly Mageia 8 net install disk for my installs.

> 
> Is your BIOS still heavily outdated as in bug 27364?

why do you say outdated ? in the 2020 case it was the latest version available with support for the Ryzen 9 5900x for the x570-A Pro mobo from the MSI support downloads, freshly flashed.

(In reply to sturmvogel from comment #35)
> Comments from duplicate bug:
> (In reply to Jean-François Bachelet from comment #5)
>  I just create partitions the
> > size I need plus one or two in sus of the usual /, /usr/local, /home and
> > /swap, usualy a /vms for all the virtual machines I create and a /save for
> > stocking some stuff. 
> > ....
> > you can try this too : let drakdisk auto partition, then just resize the
> > /home part and click on next, it ask which partition should be formated,
> 
> 
> Some important information is missing here. The partition scheme you
> describe in the first sentence is no "usual" scheme. This not possible to be
> created by any "autopartition".

in auto partition mode you have only the small grub part, /, /swap and /home created. 
I prefer to have a /usr/local separate partition.

the other two are useful for me.

BTW the number of partitions or the way it are created should not interfere with the formating and make it failing on M2 NVME...

> 
> And please clarify what you are doing as asked in my last comment as there
> is no "auto partition" mode (at least not called this way in drakdisk)
> https://doc.mageia.org/installer/8/en/content/diskPartitioning.html
> - Use Existing Partitions
> - Use Free Space
> - Use Free Space on a Windows Partition
> - Custom Disk Partitioning -> Auto allocate

as said : brand new disk --> custom disk partitioning with partitions added one by one, then add the mountpoints, then add a label, then format, then finish all the install, then do the updates if any, then quit drakdisk and after reboot cleanly the machine before booting on the new system. 

never use auto allocate or just for tests.

Might be interesting to try m9. The first alpha is available from the
mirrors such as in
http://mirror.math.princeton.edu/pub/mageia/iso/cauldron/

(In reply to sturmvogel from comment #36)

Le 04/12/2022 à 18:06, sturmvogel a écrit :
> https://bugs.mageia.org/show_bug.cgi?id=27364
>
> --- Comment #36 from sturmvogel <sturm-fr@web.de> ---
> According your provided old logs from 2020 there seems to be a messy setup on
> your side.
>
> Device          Start         End    Sectors  Size Type
> /dev/sda1        2048  5120000000 5119997953  2.4T Microsoft basic data
> /dev/sda2  5120002048  5187108864   67106817   32G Linux filesystem
> /dev/sda3  5187110912  5203492864   16381953  7.8G Linux filesystem
> /dev/sda4  5203494912  7300644864 2097149953 1000G Linux filesystem
> /dev/sda5  7300646912 11721041920 4420395009  2.1T Linux filesystem

this is not the M2 NVME disk as it was only 1 TB big 😄 this one is a 6 TB harddisk

that hold the frequently written-to partition of the windows installation (sda1-(documents and settings for the win users) precaution to not kill the ssd too fast 😉

and some mageia parts I've created after the bug to have a booting mageia install as the M2 NVME was not usable for it...

there you find /, /usr/local, /home, and a 2.1TB data partition, all ext4.


grub is installed on the efi partition of the M2 NVME.


> And your nvme (sdb) which you try to partition (not sure as you didn't tell
> exactly) seems borked:
>
> * found a dos partition table on /dev/sdb at sector 0
> * guess_geometry_from_partition_table sdb: 1936/255/63
> * is_geometry_valid_for_the_partition_table failed for (sdb1, 8192): 0,130,2 vs
> 4,0,0 with geometry 15193/64/32
> * is_geometry_valid_for_the_partition_table failed for (sdb1, 31116287):
> 1023,254,62 vs 1023,31,31 with geometry 15193/64/32
> * is_geometry_valid_for_the_partition_table failed for (sdb1, 31116287):
> 1023,254,62 vs 1023,229,20 with geometry 1936/255/63
> * sdb: argh! no valid geometry found
>
sdb there is a 16 Gb SDCard with Mageia 8b1 install dvd copied on it using  isodumper

that I used to boot the machine and make the installation 😄

if it seems bizzarely done please blame isodumper and the original Mageia iso not me 😉

before you ask, yes I check the md5sum of an iso before using it 😄


The NVME is named NVME0 there.

If you are still seeing the kernel bug, then this is still a kernel/driver problem. If not, please provide a new installer bug report using the 'bug' command.

Please answer comment 39 and comment 41

Status: REOPENED => NEEDINFO

>this one is a 6 TB harddisk
How much did you pay for it?
I saw someone that got a fantastic big NVME storage for a fantastic low price. However, this was a fantastic lie.

CC: (none) => yves.brungard_mageia

No response from reporter in opposite to the claim of duplicate bug 31217 "this time the machine is mine so we'll have the time to chase that bug to the bone"

Closing again.

Resolution: (none) => OLD
Status: NEEDINFO => RESOLVED