Hi Greg, Sasha,
Please consider commit ba1b40ed0e34 ("drm: nova: depend on
CONFIG_64BIT") for 6.18.y. It should cherry-pick cleanly.
Without this commit, one can create a config where `CONFIG_DRM_NOVA`
selects `CONFIG_NOVA_CORE` without satisfying its `CONFIG_64BIT`
dependency.
In turn, this means arm32 builds can fail -- Kconfig warns:
WARNING: unmet direct dependencies detected for NOVA_CORE
Depends on [n]: HAS_IOMEM [=y] && 64BIT && PCI [=y] && RUST [=y]
&& RUST_FW_LOADER_ABSTRACTIONS [=y]
Selected by [y]:
- DRM_NOVA [=y] && HAS_IOMEM [=y] && DRM [=y]=y [=y] && PCI [=y]
&& RUST [=y]
And then the build fails with (among others, see the related commit
5c5a41a75452 ("gpu: nova-core: depend on CONFIG_64BIT") for more):
error[E0308]: mismatched types
--> drivers/gpu/nova-core/fb.rs:50:59
|
50 | hal::fb_hal(chipset).write_sysmem_flush_page(bar,
page.dma_handle())?;
| -----------------------
^^^^^^^^^^^^^^^^^ expected `u64`, found `u32`
| |
| arguments to this method are incorrect
|
Cc'ing Danilo and Alexandre so that they can confirm they agree.
Thanks!
Cheers,
Miguel
Hi -
a Gentoo user recently found that 6.18.2 started to reproducuibly
crash when building their go toolchain [1].
Apparently the addition of "sched/fair: Forfeit vruntime on yield"
(mainline 79104becf42b) can result in the infamous NULL returned from
pick_eevdf(), which is not supposed to happen.
It turned out that the mentioned commit triggered a bug related
to the recently added proxy execution feature, which was already
fixed in mainline by "sched/proxy: Yield the donor task"
(127b90315ca0), though not marked for stable.
Applying this to 6.18.2/.3-rc1 (and probably 6.12 as well)
has reproducibly fixed the problem. A possible reason the crash
was triggered by the Go runtime could be its specific use of yield(),
though that's just speculation on my part.
So please add 127b90315ca0 ("sched/proxy: Yield the donor task")
to 6.18.y/6.12.y. I know we're already in 6.18.3-rc1, but the
crasher seems reproducible.
Fernand, please correct me if I got the explanations wrong.
Thanks!
Holger
[1] https://bugs.gentoo.org/968116 starting at #8
The reference obtained by calling usb_get_dev() is not released in the
gpio_mpsse_probe() error paths. Fix that by using device managed helper
functions. Also remove the usb_put_dev() call in the disconnect function
since now it will be released automatically.
Cc: stable(a)vger.kernel.org
Fixes: c46a74ff05c0 ("gpio: add support for FTDI's MPSSE as GPIO")
Signed-off-by: Abdun Nihaal <nihaal(a)cse.iitm.ac.in>
---
Compile tested only. Not tested on real hardware.
v1->v2:
- Switched to use devm_add_action_or_reset() to avoid unnecessary gotos,
as suggested by Bartosz Golaszewski.
Link to v1: https://lore.kernel.org/all/20251223065306.131008-1-nihaal@cse.iitm.ac.in/
drivers/gpio/gpio-mpsse.c | 12 +++++++++++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/drivers/gpio/gpio-mpsse.c b/drivers/gpio/gpio-mpsse.c
index ace652ba4df1..12191aeb6566 100644
--- a/drivers/gpio/gpio-mpsse.c
+++ b/drivers/gpio/gpio-mpsse.c
@@ -548,6 +548,13 @@ static void gpio_mpsse_ida_remove(void *data)
ida_free(&gpio_mpsse_ida, priv->id);
}
+static void gpio_mpsse_usb_put_dev(void *data)
+{
+ struct mpsse_priv *priv = data;
+
+ usb_put_dev(priv->udev);
+}
+
static int mpsse_init_valid_mask(struct gpio_chip *chip,
unsigned long *valid_mask,
unsigned int ngpios)
@@ -592,6 +599,10 @@ static int gpio_mpsse_probe(struct usb_interface *interface,
INIT_LIST_HEAD(&priv->workers);
priv->udev = usb_get_dev(interface_to_usbdev(interface));
+ err = devm_add_action_or_reset(dev, gpio_mpsse_usb_put_dev, priv);
+ if (err)
+ return err;
+
priv->intf = interface;
priv->intf_id = interface->cur_altsetting->desc.bInterfaceNumber;
@@ -713,7 +724,6 @@ static void gpio_mpsse_disconnect(struct usb_interface *intf)
priv->intf = NULL;
usb_set_intfdata(intf, NULL);
- usb_put_dev(priv->udev);
}
static struct usb_driver gpio_mpsse_driver = {
--
2.43.0
Hi and happy new year!
I would like to report a problem that I am encountering with the sdcard
storage.
I have a laptop/convertible where /home is on an sd-card (because the
internal disk is too small). The card is luks encrypted and has a btrfs
filesystem on it.
When the laptop sleeps and then resumes, there is a race. The sdcard
gets accessed for read/write but is not yet ready, so there are I/O
errors. BTRFS is not happy with them and tends to remount RO.
This issue is well known to purism developers (e.g. see
https://source.puri.sm/Librem5/linux/-/issues/484 and
https://forums.puri.sm/t/sdcard-becomes-read-only-after-waking-up-from-susp…).
My kernel logs are identical to those in
https://source.puri.sm/Librem5/linux/-/issues/484 (first comment), apart
from the fact that I get the errors from BTRFS, while the reporter there
gets the errors from EXT4. This indicates that the race is not specific
to BTRFS.
The errors in the kernel logs come right after the PM: suspend exit message.
From what I understand:
1. The error is more frequent with the SD-LUKS-FILESYSTEM
stratification, but not specific to it
2. A phone/tablet set up such as those that purism developers address
will generally use sdcard for storage and require suspend, being a good
trigger for the problem. However, the problem is in no means specific to
phones, ARM devices, etc. I am getting it on an X86-64 laptop.
3. It is unclear to me if there is a real risk of data loss. Possibly
with BTRFS that has a more complex data management this can be the case.
4.Even without data loss, the issue requires a reboot to get the
filesystem back to RW, so it is annoying.
5. Purism developers have a kernel patch for it at
https://source.puri.sm/Librem5/linux/-/merge_requests/788, but I believe
it is not in linux mainline or stable. Would it make sense to consider
that patch?
6. For stable kernels, there is a mitigation consisting in a systemd
sleep-resume hook as in
#!/bin/sh
/usr/bin/systemd-cat -p5 /usr/bin/echo ${1} ${2}
case "${1}" in
post)
sleep 1.5
systemd-cat -p4 /usr/bin/echo "hack, wait for sdcard"
;;
esac
see https://source.puri.sm/Librem5/linux/-/issues/484#note_277648
This appears to reduce the occurrence of the problem, but not to
eliminate it completely.
Thanks for the attention
Sergio
The issue occurs when gfs2_freeze_lock_shared() fails in
gfs2_fill_super(). If !sb_rdonly(sb), threads for the quotad and logd
were started, however, in the error path for gfs2_freeze_lock_shared(),
the threads are not stopped by gfs2_destroy_threads() before jumping to
fail_per_node.
Introduce fail_threads to handle stopping the threads if the threads were
started.
Reported-by: syzbot+4cb0d0336db6bc6930e9(a)syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=4cb0d0336db6bc6930e9
Fixes: a28dc123fa66 ("gfs2: init system threads before freeze lock")
Cc: stable(a)vger.kernel.org
Signed-off-by: Ryota Sakamoto <sakamo.ryota(a)gmail.com>
---
Changes in v2:
- Fix commit message style (imperative mood) as suggested by Markus Elfring.
- Add parentheses to function name in subject as suggested by Markus Elfring.
- Link to v1: https://lore.kernel.org/r/20251230-fix-use-after-free-gfs2-v1-1-ef0e46db6ec…
---
fs/gfs2/ops_fstype.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
index e7a88b717991ae3647c1da039636daef7005a7f0..4b5ac1a7050f1fd34e10be4100a2bc381f49c83d 100644
--- a/fs/gfs2/ops_fstype.c
+++ b/fs/gfs2/ops_fstype.c
@@ -1269,21 +1269,23 @@ static int gfs2_fill_super(struct super_block *sb, struct fs_context *fc)
error = gfs2_freeze_lock_shared(sdp);
if (error)
- goto fail_per_node;
+ goto fail_threads;
if (!sb_rdonly(sb))
error = gfs2_make_fs_rw(sdp);
if (error) {
gfs2_freeze_unlock(sdp);
- gfs2_destroy_threads(sdp);
fs_err(sdp, "can't make FS RW: %d\n", error);
- goto fail_per_node;
+ goto fail_threads;
}
gfs2_glock_dq_uninit(&mount_gh);
gfs2_online_uevent(sdp);
return 0;
+fail_threads:
+ if (!sb_rdonly(sb))
+ gfs2_destroy_threads(sdp);
fail_per_node:
init_per_node(sdp, UNDO);
fail_inodes:
---
base-commit: 7839932417dd53bb09eb5a585a7a92781dfd7cb2
change-id: 20251230-fix-use-after-free-gfs2-66cfbe23baa8
Best regards,
--
Ryota Sakamoto <sakamo.ryota(a)gmail.com>
Dear,
Please grant me the permission to share important discussions with you. I
am looking forward to hearing from you at your earliest convenience.
Best Regards.
Mrs. Mariam Kouame
The patch below does not apply to the 6.12-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.12.y
git checkout FETCH_HEAD
git cherry-pick -x f5e1e5ec204da11fa87fdf006d451d80ce06e118
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025122951-giggle-reveler-2e6b@gregkh' --subject-prefix 'PATCH 6.12.y' HEAD^..
Possible dependencies:
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From f5e1e5ec204da11fa87fdf006d451d80ce06e118 Mon Sep 17 00:00:00 2001
From: Tejun Heo <tj(a)kernel.org>
Date: Thu, 11 Dec 2025 15:45:04 -1000
Subject: [PATCH] sched_ext: Fix missing post-enqueue handling in
move_local_task_to_local_dsq()
move_local_task_to_local_dsq() is used when moving a task from a non-local
DSQ to a local DSQ on the same CPU. It directly manipulates the local DSQ
without going through dispatch_enqueue() and was missing the post-enqueue
handling that triggers preemption when SCX_ENQ_PREEMPT is set or the idle
task is running.
The function is used by move_task_between_dsqs() which backs
scx_bpf_dsq_move() and may be called while the CPU is busy.
Add local_dsq_post_enq() call to move_local_task_to_local_dsq(). As the
dispatch path doesn't need post-enqueue handling, add SCX_RQ_IN_BALANCE
early exit to keep consume_dispatch_q() behavior unchanged and avoid
triggering unnecessary resched when scx_bpf_dsq_move() is used from the
dispatch path.
Fixes: 4c30f5ce4f7a ("sched_ext: Implement scx_bpf_dispatch[_vtime]_from_dsq()")
Cc: stable(a)vger.kernel.org # v6.12+
Reviewed-by: Andrea Righi <arighi(a)nvidia.com>
Reviewed-by: Emil Tsalapatis <emil(a)etsalapatis.com>
Signed-off-by: Tejun Heo <tj(a)kernel.org>
diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
index c78efa99406f..695503a2f7d1 100644
--- a/kernel/sched/ext.c
+++ b/kernel/sched/ext.c
@@ -988,6 +988,14 @@ static void local_dsq_post_enq(struct scx_dispatch_q *dsq, struct task_struct *p
struct rq *rq = container_of(dsq, struct rq, scx.local_dsq);
bool preempt = false;
+ /*
+ * If @rq is in balance, the CPU is already vacant and looking for the
+ * next task to run. No need to preempt or trigger resched after moving
+ * @p into its local DSQ.
+ */
+ if (rq->scx.flags & SCX_RQ_IN_BALANCE)
+ return;
+
if ((enq_flags & SCX_ENQ_PREEMPT) && p != rq->curr &&
rq->curr->sched_class == &ext_sched_class) {
rq->curr->scx.slice = 0;
@@ -1636,6 +1644,8 @@ static void move_local_task_to_local_dsq(struct task_struct *p, u64 enq_flags,
dsq_mod_nr(dst_dsq, 1);
p->scx.dsq = dst_dsq;
+
+ local_dsq_post_enq(dst_dsq, p, enq_flags);
}
/**