Re: BUG: net-next (7.0-rc6 based and later) fails to boot on Jetson Xavier NX

Re: BUG: net-next (7.0-rc6 based and later) fails to boot on Jetson Xavier NX

[Russell King (Oracle)]

On Wed, Apr 08, 2026 at 02:59:42PM +0100, Russell King (Oracle) wrote:
> On Wed, Apr 08, 2026 at 02:07:36PM +0100, Russell King (Oracle) wrote:
> > Hi,
> > 
> > Just a heads-up that current net-next (v7.0-rc6 based) fails to boot on
> > my nVidia Jetson Xavier platform. v7.0-rc5 and v6.14 based net-next both
> > boot fine. This is an arm64 platform.
> > 
> > The problem appears to be completely random in terms of its symptoms,
> > and looks like severe memory corruption - every boot seems to produce
> > a different problem. The common theme is, although the kernel gets to
> > userspace, it never gets anywhere close to a login prompt before
> > failing in some way.
> > 
> > The last net-next+ boot (which is currently v7.0-rc6 based) resulted
> > in:
> > 
> > tegra-mc 2c00000.memory-controller: xusb_hostw: secure write @0x00000003ffffff00: VPR violation ((null))
> > ...
> > irq 91: nobody cared (try booting with the "irqpoll" option)
> > ...
> > depmod: ERROR: could not open directory /lib/modules/7.0.0-rc6-net-next+: No such file or directory
> > ...
> > Unable to handle kernel paging request at virtual address 0003201fd50320cf
> > 
> > 
> > A previous boot of the exact same kernel didn't oops, but was unable
> > to find the block device to mount for /mnt via block UUID.
> > 
> > A previous boot to that resulted in an oops.
> > 
> > 
> > The intersting thing is - the depmod error above is incorrect:
> > 
> > root@tegra-ubuntu:~# ls -ld /lib/modules/7.0.0-rc6-net-next+
> > drwxrwxr-x 3 root root 4096 Apr  8 10:23 /lib/modules/7.0.0-rc6-net-next+
> > 
> > The directory is definitely there, and is readable - checked after
> > booting back into net-next based on 7.0-rc5. In some of these boots,
> > stmmac hasn't probed yet, which rules out my changes.
> > 
> > Rootfs is ext4, and it seems there were a lot of ext4 commits merged
> > between rc5 and rc6, but nothing for rc7.
> > 
> > My current net-next head is dfecb0c5af3b. Merging rc7 on top also
> > fails, I suspect also randomly, with that I just got:
> > 
> > EXT4-fs (mmcblk0p1): VFS: Can't find ext4 filesystem
> > mount: /mnt: wrong fs type, bad option, bad superblock on /dev/mmcblk0p1, missing codepage or helper program, or other error.
> > mount: /mnt/: can't find PARTUUID=741c0777-391a-4bce-a222-455e180ece2a.
> > Unable to handle kernel paging request at virtual address f9bf0011ac0fb893
> > Mem abort info:
> >   ESR = 0x0000000096000004
> >   EC = 0x25: DABT (current EL), IL = 32 bits
> >   SET = 0, FnV = 0
> >   EA = 0, S1PTW = 0
> >   FSC = 0x04: level 0 translation fault
> > Data abort info:
> >   ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
> >   CM = 0, WnR = 0, TnD = 0, TagAccess = 0
> >   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
> > [f9bf0011ac0fb893] address between user and kernel address ranges
> > Internal error: Oops: 0000000096000004 [#1]  SMP
> > Modules linked in:
> > CPU: 1 UID: 0 PID: 936 Comm: mount Not tainted 7.0.0-rc7-net-next+ #649 PREEMPT
> > Hardware name: NVIDIA NVIDIA Jetson Xavier NX Developer Kit/Jetson, BIOS 6.0-37391689 08/28/2024
> > pstate: 20400009 (nzCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> > pc : refill_objects+0x298/0x5ec
> > lr : refill_objects+0x1f0/0x5ec
> > 
> > ...
> > 
> > Call trace:
> >  refill_objects+0x298/0x5ec (P)
> >  __pcs_replace_empty_main+0x13c/0x3a8
> >  kmem_cache_alloc_noprof+0x324/0x3a0
> >  alloc_iova+0x3c/0x290
> >  alloc_iova_fast+0x168/0x2d4
> >  iommu_dma_alloc_iova+0x84/0x154
> >  iommu_dma_map_sg+0x2c4/0x538
> >  __dma_map_sg_attrs+0x124/0x2c0
> >  dma_map_sg_attrs+0x10/0x20
> >  sdhci_pre_dma_transfer+0xb8/0x164
> >  sdhci_pre_req+0x38/0x44
> >  mmc_blk_mq_issue_rq+0x3dc/0x920
> >  mmc_mq_queue_rq+0x104/0x2b0
> >  __blk_mq_issue_directly+0x38/0xb0
> >  blk_mq_request_issue_directly+0x54/0xb4
> >  blk_mq_issue_direct+0x84/0x180
> >  blk_mq_dispatch_queue_requests+0x1a8/0x2e0
> >  blk_mq_flush_plug_list+0x60/0x140
> >  __blk_flush_plug+0xe0/0x11c
> >  blk_finish_plug+0x38/0x4c
> >  read_pages+0x158/0x260
> >  page_cache_ra_unbounded+0x158/0x3e0
> >  force_page_cache_ra+0xb0/0xe4
> >  page_cache_sync_ra+0x88/0x480
> >  filemap_get_pages+0xd8/0x850
> >  filemap_read+0xdc/0x3d8
> >  blkdev_read_iter+0x84/0x198
> >  vfs_read+0x208/0x2d8
> >  ksys_read+0x58/0xf4
> >  __arm64_sys_read+0x1c/0x28
> >  invoke_syscall.constprop.0+0x50/0xe0
> >  do_el0_svc+0x40/0xc0
> >  el0_svc+0x48/0x2a0
> >  el0t_64_sync_handler+0xa0/0xe4
> >  el0t_64_sync+0x19c/0x1a0
> > Code: 54000189 f9000022 aa0203e4 b9402ae3 (f8634840)
> > ---[ end trace 0000000000000000 ]---
> > Kernel panic - not syncing: Oops: Fatal exception
> > 
> > Looking at the changes between rc5 and rc6, there's one drivers/block
> > change for zram (which is used on this platform), one change in
> > drivers/base for regmap, nothing for drivers/mmc, but plenty for
> > fs/ext4. There are five DMA API changes.
> > 
> > Now building straight -rc7. If that also fails, my plan is to start
> > bisecting rc5..rc6, which will likely take most of the rest of the
> > day. So, in the mean time I'm sending this as a heads-up that rc6
> > and onwards has a problem.
> 
> Plain -rc7 fails (another random oops):
> 
> Root device found: PARTUUID=741c0777-391a-4bce-a222-455e180ece2a
> depmod: ERROR: could not open directory /lib/modules/7.0.0-rc7-net-next+: No such file or directory
> depmod: FATAL: could not search modules: No such file or directory
> usb 2-3: new SuperSpeed Plus Gen 2x1 USB device number 2 using tegra-xusb
> hub 2-3:1.0: USB hub found
> hub 2-3:1.0: 4 ports detected
> usb 1-3: new full-speed USB device number 3 using tegra-xusb
> Unable to handle kernel paging request at virtual address 0003201fd50320cf
> Mem abort info:
>   ESR = 0x0000000096000004
>   EC = 0x25: DABT (current EL), IL = 32 bits
>   SET = 0, FnV = 0
>   EA = 0, S1PTW = 0
>   FSC = 0x04: level 0 translation fault
> Data abort info:
>   ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
>   CM = 0, WnR = 0, TnD = 0, TagAccess = 0
>   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
> [0003201fd50320cf] address between user and kernel address ranges
> Internal error: Oops: 0000000096000004 [#1]  SMP
> Modules linked in:
> CPU: 1 UID: 0 PID: 917 Comm: mount Not tainted 7.0.0-rc7-net-next+ #649 PREEMPT
> Hardware name: NVIDIA NVIDIA Jetson Xavier NX Developer Kit/Jetson, BIOS 6.0-37391689 08/28/2024
> pstate: 20400009 (nzCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> pc : refill_objects+0x298/0x5ec
> lr : refill_objects+0x1f0/0x5ec
> sp : ffff80008606b500
> x29: ffff80008606b500 x28: 0000000000000001 x27: fffffdffc20e6200
> x26: 0000000000000006 x25: 0000000000000000 x24: 000000000000003c
> x23: ffff0000809e4840 x22: ffff0000809dba00 x21: ffff80008606b5a0
> x20: ffff000081133820 x19: fffffdffc20e6220 x18: 0000000000000000
> x17: 0000000000000000 x16: 0000000000000100 x15: 0000000000000000
> x14: 0000000000000000 x13: 0000000000000000 x12: ffff800081e5faa8
> x11: ffff800082192c70 x10: ffff8000814074dc x9 : 0000000000000050
> x8 : ffff80008606b490 x7 : ffff000083988b40 x6 : ffff80008606b4a0
> x5 : 000000080015000f x4 : d503201fd503201f x3 : 00000000000000b0
> x2 : d503201fd503201f x1 : ffff000081133828 x0 : d503201fd503201f
> Call trace:
>  refill_objects+0x298/0x5ec (P)
>  __pcs_replace_empty_main+0x13c/0x3a8
>  kmem_cache_alloc_noprof+0x324/0x3a0
>  mempool_alloc_slab+0x1c/0x28
>  mempool_alloc_noprof+0x98/0xe0
>  bio_alloc_bioset+0x160/0x3e0
>  do_mpage_readpage+0x3d0/0x618
>  mpage_readahead+0xb8/0x144
>  blkdev_readahead+0x18/0x24
>  read_pages+0x58/0x260
>  page_cache_ra_unbounded+0x158/0x3e0
>  force_page_cache_ra+0xb0/0xe4
>  page_cache_sync_ra+0x88/0x480
>  filemap_get_pages+0xd8/0x850
>  filemap_read+0xdc/0x3d8
>  blkdev_read_iter+0x84/0x198
>  vfs_read+0x208/0x2d8
>  ksys_read+0x58/0xf4
>  __arm64_sys_read+0x1c/0x28
>  invoke_syscall.constprop.0+0x50/0xe0
>  do_el0_svc+0x40/0xc0
>  el0_svc+0x48/0x2a0
>  el0t_64_sync_handler+0xa0/0xe4
>  el0t_64_sync+0x19c/0x1a0
> Code: 54000189 f9000022 aa0203e4 b9402ae3 (f8634840)
> ---[ end trace 0000000000000000 ]---
> 
> Now starting the bisect between 7.0-rc5 and 7.0-rc6.

The rebase is still progressing, but it's landed on:

c7d812e33f3e dmaengine: xilinx: xilinx_dma: Fix unmasked residue subtraction

and while this boots to a login prompt, it spat out a BUG():

BUG: sleeping function called from invalid context at kernel/locking/mutex.c:591
in_atomic(): 0, irqs_disabled(): 1, non_block: 0, pid: 56, name: kworker/u24:3
preempt_count: 0, expected: 0
RCU nest depth: 0, expected: 0
3 locks held by kworker/u24:3/56:
 #0: ffff000080042148 ((wq_completion)events_unbound#2){+.+.}-{0:0}, at: process_one_work+0x184/0x780
 #1: ffff80008299bdf8 (deferred_probe_work){+.+.}-{0:0}, at: process_one_work+0x1ac/0x780
 #2: ffff0000808b48f8 (&dev->mutex){....}-{4:4}, at: __device_attach+0x2c/0x188
irq event stamp: 10872
hardirqs last  enabled at (10871): [<ffff80008013a410>] ktime_get+0x130/0x180
hardirqs last disabled at (10872): [<ffff800080d61ac8>] _raw_spin_lock_irqsave+0x84/0x88
softirqs last  enabled at (9216): [<ffff80008002807c>] fpsimd_save_and_flush_current_state+0x3c/0x80
softirqs last disabled at (9214): [<ffff800080028098>] fpsimd_save_and_flush_current_state+0x58/0x80
CPU: 5 UID: 0 PID: 56 Comm: kworker/u24:3 Not tainted 7.0.0-rc1-bisect+ #654 PREEMPT
Hardware name: NVIDIA NVIDIA Jetson Xavier NX Developer Kit/Jetson, BIOS 6.0-37391689 08/28/2024
Workqueue: events_unbound deferred_probe_work_func
Call trace:
 show_stack+0x18/0x30 (C)
 dump_stack_lvl+0x6c/0x94
 dump_stack+0x18/0x24
 __might_resched+0x154/0x220
 __might_sleep+0x48/0x80
 __mutex_lock+0x48/0x800
 mutex_lock_nested+0x24/0x30
 pinmux_disable_setting+0x9c/0x180
 pinctrl_commit_state+0x5c/0x260
 pinctrl_pm_select_idle_state+0x4c/0xa0
 tegra_i2c_runtime_suspend+0x2c/0x3c
 pm_generic_runtime_suspend+0x2c/0x44
 __rpm_callback+0x48/0x1ec
 rpm_callback+0x74/0x80
 rpm_suspend+0xec/0x630
 rpm_idle+0x2c0/0x420
 __pm_runtime_idle+0x44/0x160
 tegra_i2c_probe+0x2e4/0x640
 platform_probe+0x5c/0xa4
 really_probe+0xbc/0x2c0
 __driver_probe_device+0x78/0x120
 driver_probe_device+0x3c/0x160
 __device_attach_driver+0xbc/0x160
 bus_for_each_drv+0x70/0xb8
 __device_attach+0xa4/0x188
 device_initial_probe+0x50/0x54
 bus_probe_device+0x38/0xa4
 deferred_probe_work_func+0x90/0xcc
 process_one_work+0x204/0x780
 worker_thread+0x1c8/0x36c
 kthread+0x138/0x144
 ret_from_fork+0x10/0x20

This is reproducible.

Adding Vinod and Frank, and dmaengine mailing list.

Bisect continuing, assuming this is a "good" commit as it isn't
producing the boot failure with random memory corruption.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: BUG: net-next (7.0-rc6 based and later) fails to boot on Jetson Xavier NX

[Russell King (Oracle)]

On Wed, Apr 08, 2026 at 05:08:34PM +0100, Russell King (Oracle) wrote:
> The rebase is still progressing, but it's landed on:
> 
> c7d812e33f3e dmaengine: xilinx: xilinx_dma: Fix unmasked residue subtraction
> 
> and while this boots to a login prompt, it spat out a BUG():
> 
> BUG: sleeping function called from invalid context at kernel/locking/mutex.c:591
> in_atomic(): 0, irqs_disabled(): 1, non_block: 0, pid: 56, name: kworker/u24:3
> preempt_count: 0, expected: 0
> RCU nest depth: 0, expected: 0
> 3 locks held by kworker/u24:3/56:
>  #0: ffff000080042148 ((wq_completion)events_unbound#2){+.+.}-{0:0}, at: process_one_work+0x184/0x780
>  #1: ffff80008299bdf8 (deferred_probe_work){+.+.}-{0:0}, at: process_one_work+0x1ac/0x780
>  #2: ffff0000808b48f8 (&dev->mutex){....}-{4:4}, at: __device_attach+0x2c/0x188
> irq event stamp: 10872
> hardirqs last  enabled at (10871): [<ffff80008013a410>] ktime_get+0x130/0x180
> hardirqs last disabled at (10872): [<ffff800080d61ac8>] _raw_spin_lock_irqsave+0x84/0x88
> softirqs last  enabled at (9216): [<ffff80008002807c>] fpsimd_save_and_flush_current_state+0x3c/0x80
> softirqs last disabled at (9214): [<ffff800080028098>] fpsimd_save_and_flush_current_state+0x58/0x80
> CPU: 5 UID: 0 PID: 56 Comm: kworker/u24:3 Not tainted 7.0.0-rc1-bisect+ #654 PREEMPT
> Hardware name: NVIDIA NVIDIA Jetson Xavier NX Developer Kit/Jetson, BIOS 6.0-37391689 08/28/2024
> Workqueue: events_unbound deferred_probe_work_func
> Call trace:
>  show_stack+0x18/0x30 (C)
>  dump_stack_lvl+0x6c/0x94
>  dump_stack+0x18/0x24
>  __might_resched+0x154/0x220
>  __might_sleep+0x48/0x80
>  __mutex_lock+0x48/0x800
>  mutex_lock_nested+0x24/0x30
>  pinmux_disable_setting+0x9c/0x180
>  pinctrl_commit_state+0x5c/0x260
>  pinctrl_pm_select_idle_state+0x4c/0xa0
>  tegra_i2c_runtime_suspend+0x2c/0x3c
>  pm_generic_runtime_suspend+0x2c/0x44
>  __rpm_callback+0x48/0x1ec
>  rpm_callback+0x74/0x80
>  rpm_suspend+0xec/0x630
>  rpm_idle+0x2c0/0x420
>  __pm_runtime_idle+0x44/0x160
>  tegra_i2c_probe+0x2e4/0x640
>  platform_probe+0x5c/0xa4
>  really_probe+0xbc/0x2c0
>  __driver_probe_device+0x78/0x120
>  driver_probe_device+0x3c/0x160
>  __device_attach_driver+0xbc/0x160
>  bus_for_each_drv+0x70/0xb8
>  __device_attach+0xa4/0x188
>  device_initial_probe+0x50/0x54
>  bus_probe_device+0x38/0xa4
>  deferred_probe_work_func+0x90/0xcc
>  process_one_work+0x204/0x780
>  worker_thread+0x1c8/0x36c
>  kthread+0x138/0x144
>  ret_from_fork+0x10/0x20
> 
> This is reproducible.

I've just realised that it's the Tegra I2C bug that is already known
about, but took ages to be fixed in mainline - it's unrelated to the
memory corruption, so can be ignored. Sorry for the noise.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: BUG: net-next (7.0-rc6 based and later) fails to boot on Jetson Xavier NX

[Robin Murphy]

On 2026-04-08 5:16 pm, Russell King (Oracle) wrote:
> On Wed, Apr 08, 2026 at 05:08:34PM +0100, Russell King (Oracle) wrote:
>> The rebase is still progressing, but it's landed on:
>>
>> c7d812e33f3e dmaengine: xilinx: xilinx_dma: Fix unmasked residue subtraction

FWIW I don't see a Tegra having the Xilinx IP in it anyway - judging by 
the DT it has their own tegra-gpcdma engine...

There's a fair chance this could be 90c5def10bea ("iommu: Do not call 
drivers for empty gathers"), which JonH also reported causing boot 
issues on Tegras - in short, SMMU TLB maintenance may not be completed 
properly which could lead to recycled DMA addresses causing exactly this 
kind of random memory corruption. I CC'd you on a patch:

https://lore.kernel.org/linux-iommu/20260408162846.GE3357077@nvidia.com/T/#t

Thanks,
Robin.

>>
>> and while this boots to a login prompt, it spat out a BUG():
>>
>> BUG: sleeping function called from invalid context at kernel/locking/mutex.c:591
>> in_atomic(): 0, irqs_disabled(): 1, non_block: 0, pid: 56, name: kworker/u24:3
>> preempt_count: 0, expected: 0
>> RCU nest depth: 0, expected: 0
>> 3 locks held by kworker/u24:3/56:
>>   #0: ffff000080042148 ((wq_completion)events_unbound#2){+.+.}-{0:0}, at: process_one_work+0x184/0x780
>>   #1: ffff80008299bdf8 (deferred_probe_work){+.+.}-{0:0}, at: process_one_work+0x1ac/0x780
>>   #2: ffff0000808b48f8 (&dev->mutex){....}-{4:4}, at: __device_attach+0x2c/0x188
>> irq event stamp: 10872
>> hardirqs last  enabled at (10871): [<ffff80008013a410>] ktime_get+0x130/0x180
>> hardirqs last disabled at (10872): [<ffff800080d61ac8>] _raw_spin_lock_irqsave+0x84/0x88
>> softirqs last  enabled at (9216): [<ffff80008002807c>] fpsimd_save_and_flush_current_state+0x3c/0x80
>> softirqs last disabled at (9214): [<ffff800080028098>] fpsimd_save_and_flush_current_state+0x58/0x80
>> CPU: 5 UID: 0 PID: 56 Comm: kworker/u24:3 Not tainted 7.0.0-rc1-bisect+ #654 PREEMPT
>> Hardware name: NVIDIA NVIDIA Jetson Xavier NX Developer Kit/Jetson, BIOS 6.0-37391689 08/28/2024
>> Workqueue: events_unbound deferred_probe_work_func
>> Call trace:
>>   show_stack+0x18/0x30 (C)
>>   dump_stack_lvl+0x6c/0x94
>>   dump_stack+0x18/0x24
>>   __might_resched+0x154/0x220
>>   __might_sleep+0x48/0x80
>>   __mutex_lock+0x48/0x800
>>   mutex_lock_nested+0x24/0x30
>>   pinmux_disable_setting+0x9c/0x180
>>   pinctrl_commit_state+0x5c/0x260
>>   pinctrl_pm_select_idle_state+0x4c/0xa0
>>   tegra_i2c_runtime_suspend+0x2c/0x3c
>>   pm_generic_runtime_suspend+0x2c/0x44
>>   __rpm_callback+0x48/0x1ec
>>   rpm_callback+0x74/0x80
>>   rpm_suspend+0xec/0x630
>>   rpm_idle+0x2c0/0x420
>>   __pm_runtime_idle+0x44/0x160
>>   tegra_i2c_probe+0x2e4/0x640
>>   platform_probe+0x5c/0xa4
>>   really_probe+0xbc/0x2c0
>>   __driver_probe_device+0x78/0x120
>>   driver_probe_device+0x3c/0x160
>>   __device_attach_driver+0xbc/0x160
>>   bus_for_each_drv+0x70/0xb8
>>   __device_attach+0xa4/0x188
>>   device_initial_probe+0x50/0x54
>>   bus_probe_device+0x38/0xa4
>>   deferred_probe_work_func+0x90/0xcc
>>   process_one_work+0x204/0x780
>>   worker_thread+0x1c8/0x36c
>>   kthread+0x138/0x144
>>   ret_from_fork+0x10/0x20
>>
>> This is reproducible.
> 
> I've just realised that it's the Tegra I2C bug that is already known
> about, but took ages to be fixed in mainline - it's unrelated to the
> memory corruption, so can be ignored. Sorry for the noise.
> 

Re: BUG: net-next (7.0-rc6 based and later) fails to boot on Jetson Xavier NX

[Russell King (Oracle)]

On Wed, Apr 08, 2026 at 05:40:48PM +0100, Robin Murphy wrote:
> On 2026-04-08 5:16 pm, Russell King (Oracle) wrote:
> > On Wed, Apr 08, 2026 at 05:08:34PM +0100, Russell King (Oracle) wrote:
> > > The rebase is still progressing, but it's landed on:
> > > 
> > > c7d812e33f3e dmaengine: xilinx: xilinx_dma: Fix unmasked residue subtraction
> 
> FWIW I don't see a Tegra having the Xilinx IP in it anyway - judging by the
> DT it has their own tegra-gpcdma engine...
> 
> There's a fair chance this could be 90c5def10bea ("iommu: Do not call
> drivers for empty gathers"), which JonH also reported causing boot issues on
> Tegras - in short, SMMU TLB maintenance may not be completed properly which
> could lead to recycled DMA addresses causing exactly this kind of random
> memory corruption. I CC'd you on a patch:
> 
> https://lore.kernel.org/linux-iommu/20260408162846.GE3357077@nvidia.com/T/#t

Okay, bisect complete, and... no idea. It seems to suggest that 7.0-rc6
is actually fine - it ended up blaming Linus' tagging of 7.0-rc6 which
only changed the makefile. So, my assumption that because rc6 was merged
into net-next last Thursday which fails, net-next+rc7 fails, rc7 also
fails, that rc6 would also fail seems to be false.

Right, rc7 built with the same .config that rc6 was built with
definitely fails, this time with:

Root device found: PARTUUID=741c0777-391a-4bce-a222-455e180ece2a
depmod: ERROR: could not open directory /lib/modules/7.0.0-rc7-bisect: No such file or directory
depmod: FATAL: could not search modules: No such file or directory
usb 2-3: new SuperSpeed Plus Gen 2x1 USB device number 2 using tegra-xusb
hub 2-3:1.0: USB hub found
hub 2-3:1.0: 4 ports detected
usb 1-3: new full-speed USB device number 3 using tegra-xusb
EXT4-fs (mmcblk0p1): VFS: Can't find ext4 filesystem
mount: /mnt: wrong fs type, bad option, bad superblock on /dev/mmcblk0p1, missing codepage or helper program, or other error.
mount: /mnt/: can't find PARTUUID=741c0777-391a-4bce-a222-455e180ece2a.
get_swap_device: Bad swap file entry 1800c00008
get_swap_device: Bad swap file entry 1800c00008
get_swap_device: Bad swap file entry 1800c00008

So, it seems rc6 -> rc7 => fails
net-next with rc5 -> net-next with rc6 => fails

However, before I test anything else, I've just built the same rc7
which failed above with your patch applied - and that boots fine.

Now, each Thursday, net-next gets updated as that's the day that the
net tree gets sent for merging into mainline. This causes net-next's
version to increase. So something in current net-next plus in rc7 is
causing this problem.

The commit you claim needs fixing is:

$ git describe --contains 90c5def10bea
v7.0-rc7~29^2~2

which I had assumed wouldn't be in net-next.

Now, mainline had this on Thursday:

commit f8f5627a8aeab15183eef8930bf75ba88a51622f
Merge: 4c2c526b5adf ec7067e66119
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Thu Apr 2 09:57:06 2026 -0700

    Merge tag 'net-7.0-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

commit 4c2c526b5adfb580bd95316bf179327d5ee26da8
Merge: 2ec9074b28a0 8b72aa5704c7
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Thu Apr 2 09:53:16 2026 -0700

    Merge tag 'iommu-fixes-v7.0-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/iommu/linux

and merging iommu-fixes-v7.0-rc6 introduced the buggy 90c5def10bea
commit into -rc7.

However, as soon as Linus merged net-7.0-rc7, netdev maintainers merged
that exact commit back into net-next:

commit 8ffb33d7709b59ff60560f48960a73bd8a55be95
Merge: 269389ba5398 f8f5627a8aea
Author: Jakub Kicinski <kuba@kernel.org>
Date:   Thu Apr 2 10:57:09 2026 -0700

    Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Thereby bringing in that buggy commit into net-next, but with net-next
identifying itself as 7.0-rc6.

That's... confusing, but explains why current net-next which reports
itself as 7.0-rc6 _and_ rc7 both fail, but rc6 itself does not. It
also means I've wasted an entire afternoon running a useless bisect
between rc5 and rc6 due to the version numbers in net-next being
meaningless.

What's the status on the iommu fix? Is it merged into mainline yet?
If it isn't already, that means net-next remains unbootable going
into the merge window without manually carrying the fix locally.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: BUG: net-next (7.0-rc6 based and later) fails to boot on Jetson Xavier NX

[Will Deacon]

On Wed, Apr 08, 2026 at 08:52:32PM +0100, Russell King (Oracle) wrote:
> What's the status on the iommu fix? Is it merged into mainline yet?
> If it isn't already, that means net-next remains unbootable going
> into the merge window without manually carrying the fix locally.

I'll pick it up for 7.0 in the iommu tree.

Will

Re: BUG: net-next (7.0-rc6 based and later) fails to boot on Jetson Xavier NX

[Linus Torvalds]

On Wed, 8 Apr 2026 at 09:08, Russell King (Oracle)
<linux@armlinux.org.uk> wrote:
>
> The rebase is still progressing, but it's landed on:
>
> c7d812e33f3e dmaengine: xilinx: xilinx_dma: Fix unmasked residue subtraction

Well, that commit looks completely bogus.

The explanation is just garbage: when subtracting two values that may
have random crud in the top bits, it's actually likely *better* to do
the masking *after* the subtraction.

The subtract of bogus upper bits will only affect upper bits. The
carry-chain only works upwards, not downwards.

So the old code that did

                       residue += (cdma_hw->control - cdma_hw->status) &
                                  chan->xdev->max_buffer_len;

would correctly mask out the upper bits, and the result of the
subtraction would be done "modulo mac_buffer_len". Which is rather
reasonable.

The code was changed to

                       residue += (cdma_hw->control &
chan->xdev->max_buffer_len) -
                                  (cdma_hw->status &
chan->xdev->max_buffer_len);

and now it does obviously still mask out the upper bits on each of the
values), but then the subtraction is done "modulo the arithmetic C
type" (which is 'u32')

In particular, if the status bits are bigger than the control bits,
that residue addition will now add a *huge* 32-bit number. It used to
add a number that was limited by the  max_buffer_len mask.

So the "interference from those top bits" stated in the commit message
is simply NOT TRUE. It's just complete rambling garbage.

Instead, the commit purely changes the final modulus of the
subtraction - which has nothing to do with any upper bits, and
everything to do with what kind of answer you want.

I think that commit is just very very wrong. At least the commit
message is wrong. And see above why I think the changed arithmetic is
likely wrong too.

It's very possible that the 'residue' is now a random 32-bit number
with the high bits set, and you get DMA corruption.

That would explain why this happens on Jetson but I haven't seen other reports.

                    Linus