The 7.0.12 kernel is unusable (except via SSH) due to a NULL pointer
dereference in nouveau. The first one:

Jun 12 11:19:16 disset kernel: Lockdown: Xorg: raw io port access is restricted; see man kernel_lockdown.7
Jun 12 11:19:16 disset kernel: BUG: kernel NULL pointer dereference, address: 0000000000000000
Jun 12 11:19:16 disset kernel: #PF: supervisor read access in kernel mode
Jun 12 11:19:16 disset kernel: #PF: error_code(0x0000) - not-present page
Jun 12 11:19:16 disset kernel: PGD 11697a067 P4D 0
Jun 12 11:19:16 disset kernel: Oops: Oops: 0000 [#1] SMP NOPTI
Jun 12 11:19:16 disset kernel: CPU: 2 UID: 0 PID: 1226 Comm: Xorg Tainted: G S                  7.0.12+deb14-amd64 #1 PREEMPT(lazy)  Debian 7.0.12-1
Jun 12 11:19:16 disset kernel: Tainted: [S]=CPU_OUT_OF_SPEC
Jun 12 11:19:16 disset kernel: Hardware name: Dell Inc. Precision 5860 Tower/0F4Y1M, BIOS 2.7.0 12/17/2024
Jun 12 11:19:16 disset kernel: RIP: 0010:nvkm_gr_units+0x9/0x30 [nouveau]
Jun 12 11:19:16 disset kernel: Code: 74 06 ff e0 cc cc cc cc 31 c0 c3 cc cc cc cc 66 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f 1f 40 d6 0f 1f 44 00 00 <48> 8b 07 48 8b 40 58 48 85 c0 74 06 ff e0 cc cc cc cc 31 c0 c3 cc
Jun 12 11:19:16 disset kernel: RSP: 0018:ff7cc66fc603fb90 EFLAGS: 00010246
Jun 12 11:19:16 disset kernel: RAX: ff240d5cc466c800 RBX: 0000000000000020 RCX: ff240d5d012c8000
Jun 12 11:19:16 disset kernel: RDX: ff240d5ce4413400 RSI: ff7cc66fc603fc70 RDI: 0000000000000000
Jun 12 11:19:16 disset kernel: RBP: ff240d5d232e0600 R08: 000000000000000d R09: ff7cc66fc603fc70
Jun 12 11:19:16 disset kernel: R10: 0000000000000010 R11: 0000000000000000 R12: ff240d5cc466c800
Jun 12 11:19:16 disset kernel: R13: ffffffffc0d5dac0 R14: ff7cc66fc603fc70 R15: 0000000000000040
Jun 12 11:19:16 disset kernel: FS:  00007f0957dd7b80(0000) GS:ff240d64911fc000(0000) knlGS:0000000000000000
Jun 12 11:19:16 disset kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jun 12 11:19:16 disset kernel: CR2: 0000000000000000 CR3: 00000001428de003 CR4: 0000000000f73ef0
Jun 12 11:19:16 disset kernel: PKRU: 55555554
Jun 12 11:19:16 disset kernel: Call Trace:
Jun 12 11:19:16 disset kernel:  <TASK>
Jun 12 11:19:16 disset kernel:  nouveau_abi16_ioctl_getparam+0x2b5/0x3b0 [nouveau]
Jun 12 11:19:16 disset kernel:  ? drm_dev_enter+0x1d/0x60 [drm]
Jun 12 11:19:16 disset kernel:  drm_ioctl_kernel+0xae/0x100 [drm]
Jun 12 11:19:16 disset kernel:  drm_ioctl+0x2a8/0x550 [drm]
Jun 12 11:19:16 disset kernel:  ? __pfx_nouveau_abi16_ioctl_getparam+0x10/0x10 [nouveau]
Jun 12 11:19:16 disset kernel:  nouveau_drm_ioctl+0x58/0xc0 [nouveau]
Jun 12 11:19:16 disset kernel:  __x64_sys_ioctl+0x97/0xe0
Jun 12 11:19:16 disset kernel:  do_syscall_64+0xcd/0x15f0
Jun 12 11:19:16 disset kernel:  ? ktime_get_mono_fast_ns+0x35/0xe0
Jun 12 11:19:16 disset kernel:  ? nouveau_drm_ioctl+0x89/0xc0 [nouveau]
Jun 12 11:19:16 disset kernel:  ? __x64_sys_ioctl+0x97/0xe0
Jun 12 11:19:16 disset kernel:  ? do_syscall_64+0x10a/0x15f0
Jun 12 11:19:16 disset kernel:  ? handle_mm_fault+0x1e1/0x2f0
Jun 12 11:19:16 disset kernel:  ? do_user_addr_fault+0x2b4/0x7b0
Jun 12 11:19:16 disset kernel:  ? irqentry_exit+0x78/0x680
Jun 12 11:19:16 disset kernel:  ? exc_page_fault+0x82/0x1d0
Jun 12 11:19:16 disset kernel:  entry_SYSCALL_64_after_hwframe+0x76/0x7e
Jun 12 11:19:16 disset kernel: RIP: 0033:0x7f095815dd3b
Jun 12 11:19:16 disset kernel: Code: 00 48 89 44 24 18 31 c0 48 8d 44 24 60 c7 04 24 10 00 00 00 48 89 44 24 08 48 8d 44 24 20 48 89 44 24 10 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 1c 48 8b 44 24 18 64 48 2b 04 25 28 00 00
Jun 12 11:19:16 disset kernel: RSP: 002b:00007ffc4f904400 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
Jun 12 11:19:16 disset kernel: RAX: ffffffffffffffda RBX: 000055ce737fae80 RCX: 00007f095815dd3b
Jun 12 11:19:16 disset kernel: RDX: 00007ffc4f9044b0 RSI: 00000000c0106440 RDI: 0000000000000012
Jun 12 11:19:16 disset kernel: RBP: 00007ffc4f9044b0 R08: 0000000000001028 R09: 0000000000000000
Jun 12 11:19:16 disset kernel: R10: 0000000000000000 R11: 0000000000000246 R12: 00000000c0106440
Jun 12 11:19:16 disset kernel: R13: 0000000000000012 R14: 000055ce737fa700 R15: 0000000000000007
Jun 12 11:19:16 disset kernel:  </TASK>
Jun 12 11:19:16 disset kernel: Modules linked in: qrtr nls_ascii nls_cp437 vfat fat snd_sof_pci_intel_tgl snd_sof_pci_intel_cnl snd_sof_intel_hda_generic soundwire_intel snd_sof_intel_hda_sdw_bpt snd_sof_intel_hda_common snd_soc_hdac_hda snd_sof_intel_hda_mlink snd_sof_intel_hda soundwire_cadence intel_rapl_msr snd_sof_pci intel_rapl_common snd_sof_xtensa_dsp intel_uncore_frequency snd_sof intel_uncore_frequency_common intel_ifs i10nm_edac snd_sof_utils snd_ctl_led skx_edac_common snd_soc_acpi_intel_match nfit snd_soc_acpi_intel_sdca_quirks soundwire_generic_allocation x86_pkg_temp_thermal snd_soc_sdw_utils intel_powerclamp snd_soc_acpi snd_hda_codec_alc269 coretemp crc8 snd_hda_codec_realtek_lib soundwire_bus snd_hda_scodec_component snd_soc_sdca snd_hda_codec_nvhdmi snd_hda_codec_generic kvm_intel snd_hda_codec_hdmi snd_soc_avs snd_soc_hda_codec snd_hda_intel snd_hda_ext_core kvm snd_hda_codec snd_hda_core snd_soc_core snd_intel_dspcfg irqbypass dax_hmem snd_intel_sdw_acpi dell_wmi dell_pc snd_compress rapl cxl_acpi snd_hwdep
Jun 12 11:19:16 disset kernel:  dell_smbios snd_pcm_dmaengine intel_cstate platform_profile dell_wmi_sysman snd_pcm cxl_pmem dell_wmi_ddv pmt_telemetry dcdbas intel_uncore dell_smm_hwmon sparse_keymap firmware_attributes_class dell_wmi_descriptor wmi_bmof battery libnvdimm isst_if_mmio isst_if_mbox_pci pmt_discovery snd_timer pcspkr mei_me iaa_crypto pmt_class intel_sdsi isst_if_common snd mei soundcore joydev evdev binfmt_misc i2c_dev msr efi_pstore configfs nfnetlink efivarfs autofs4 ext4 crc16 mbcache jbd2 crc32c_cryptoapi dm_crypt dm_mod cdc_ncm cdc_ether usbnet nvme nvme_core nvme_keyring nvme_auth r8152 hkdf hid_generic mii usbhid hid nouveau mxm_wmi drm_gpuvm gpu_sched drm_ttm_helper ttm drm_exec i2c_algo_bit drm_display_helper cec rc_core drm_client_lib drm_kms_helper drm iTCO_wdt ahci intel_pmc_bxt libahci watchdog rtsx_pci_sdmmc xhci_pci libata mmc_core xhci_hcd ghash_clmulni_intel idxd atlantic video i2c_i801 scsi_mod usbcore aesni_intel intel_vsec idxd_bus e1000e rtsx_pci macsec wmi pinctrl_alderlake i2c_smbus button vmd
Jun 12 11:19:16 disset kernel:  scsi_common usb_common
Jun 12 11:19:16 disset kernel: CR2: 0000000000000000
Jun 12 11:19:16 disset kernel: ---[ end trace 0000000000000000 ]---
Jun 12 11:19:16 disset kernel: RIP: 0010:nvkm_gr_units+0x9/0x30 [nouveau]
Jun 12 11:19:16 disset kernel: Code: 74 06 ff e0 cc cc cc cc 31 c0 c3 cc cc cc cc 66 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f 1f 40 d6 0f 1f 44 00 00 <48> 8b 07 48 8b 40 58 48 85 c0 74 06 ff e0 cc cc cc cc 31 c0 c3 cc
Jun 12 11:19:16 disset kernel: RSP: 0018:ff7cc66fc603fb90 EFLAGS: 00010246
Jun 12 11:19:16 disset kernel: RAX: ff240d5cc466c800 RBX: 0000000000000020 RCX: ff240d5d012c8000
Jun 12 11:19:16 disset kernel: RDX: ff240d5ce4413400 RSI: ff7cc66fc603fc70 RDI: 0000000000000000
Jun 12 11:19:16 disset kernel: RBP: ff240d5d232e0600 R08: 000000000000000d R09: ff7cc66fc603fc70
Jun 12 11:19:16 disset kernel: R10: 0000000000000010 R11: 0000000000000000 R12: ff240d5cc466c800
Jun 12 11:19:16 disset kernel: R13: ffffffffc0d5dac0 R14: ff7cc66fc603fc70 R15: 0000000000000040
Jun 12 11:19:16 disset kernel: FS:  00007f0957dd7b80(0000) GS:ff240d64911fc000(0000) knlGS:0000000000000000
Jun 12 11:19:16 disset kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jun 12 11:19:16 disset kernel: CR2: 0000000000000000 CR3: 00000001428de003 CR4: 0000000000f73ef0
Jun 12 11:19:16 disset kernel: PKRU: 55555554
Jun 12 11:19:16 disset kernel: note: Xorg[1226] exited with irqs disabled

I've attached the full logs.

I get a black screen.

[...]

... with a small "_" cursor at the top-left.

The initrd is much smaller than the one of the 7.0.10 kernel
(which is still working fine):

Control: reassign -1 firmware-misc-nonfree 20260519-1
Control: retitle -1 firmware-misc-nonfree should recommend firmware-nvidia-graphics to avoid kernel Oops + black screen
Control: severity -1 serious

Here, using firmware-linux-free alone just like on the other machine
does not solve the problem.

Using firmware-nvidia-graphics without firmware-misc-nonfree makes
the machine unbootable just after unlocking the disk.

Installing firmware-nvidia-graphics in addition to firmware-misc-nonfree
solves the problem. So I'm reassigning the bug.

The change from "Recommends:" to "Suggests:" was due to a request in

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1137651

but this should rather be solved by the installer. BTW, the installer
currently does *not* install "Recommends:" by default, so that #1137651
wouldn't even have been solved! See

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=931283

features, but here, the machine is completely unusable without network
access. The Linux virtual consoles (Ctrl+Alt+F<1-6>) do not even work.

I'm also wondering why it is necessary to have a 100 MB firmware just
to get a basic display. In the past (or with the Nvidia proprietary
drivers?), this was not the case. But this is another issue.

Control: tag -1 moreinfo unreproducible
[...]

By default, a Suggests relationship from an explicitly installed package
blocks autoremoval.  I tested this before making the change, and I just
verified it again now:

So, did you change this default behaviour?  (I don't know how you would
- there isn't a documented setting for it in apt.conf.)

Or did you use some other tool to autoremove packages, like deborphan?

Ben.

The huge "GSP" firmware has most of the code that used to be in the
proprietary driver, now running on a microcontroller next to the GPU.
For these newer GPUs, the remaining driver code required on the host is
apparently not as sensitive, so even Nvidia's own driver is open source
(but not upstream).  Progress, eh?

But there is work ongoing to remove vendor-specific graphics drivers
(and their firmware) from the initramfs again.

Ben.

I have

APT::AutoRemove::SuggestsImportant "false";

The reason is that in general, when a Recommends is changed to
Suggests, this is because either the Recommends was actually a mistake
or the package that was recommended is no longer really necessary.
Another reason is the case where a package was installed due to some
dependency, and without the above setting, it is not removed when this
dependency disappears and another package has a Suggests on it. Thus
keeping Suggests has various drawbacks so that I had added this option
in the past to get rid of such useless packages (including a useless
broken daemon that was affecting my system: bug 901605).

Perhaps having APT::AutoRemove::SuggestsImportant set to true makes
sense on Debian/stable machines, but on a Debian/sid machine, where
one often installs and removes packages, this is just bad.

Control: severity -1 wishlist
Control: tag -1 wontfix
[...]

But then you also have to accept that there will sometimes be breakage
like this.

Ben.

This does not solve the issue that after installation, an Nvidia-based
machine will now be unbootable!