[TEST] SoundWire sysfs upstream changes by plbossart · Pull Request #4799 · thesofproject/linux

plbossart · 2024-01-31T08:08:50Z

need to check if the hierarchy is preserved.

Add a mechanism for named attribute_groups to hide their directory at sysfs_update_group() time, or otherwise skip emitting the group directory when the group is first registered. It piggybacks on is_visible() in a similar manner as SYSFS_PREALLOC, i.e. special flags in the upper bits of the returned mode. To use it, specify a symbol prefix to DEFINE_SYSFS_GROUP_VISIBLE(), and then pass that same prefix to SYSFS_GROUP_VISIBLE() when assigning the @is_visible() callback: DEFINE_SYSFS_GROUP_VISIBLE($prefix) struct attribute_group $prefix_group = { .name = $name, .is_visible = SYSFS_GROUP_VISIBLE($prefix), }; SYSFS_GROUP_VISIBLE() expects a definition of $prefix_group_visible() and $prefix_attr_visible(), where $prefix_group_visible() just returns true / false and $prefix_attr_visible() behaves as normal. The motivation for this capability is to centralize PCI device authentication in the PCI core with a named sysfs group while keeping that group hidden for devices and platforms that do not meet the requirements. In a PCI topology, most devices will not support authentication, a small subset will support just PCI CMA (Component Measurement and Authentication), a smaller subset will support PCI CMA + PCIe IDE (Link Integrity and Encryption), and only next generation server hosts will start to include a platform TSM (TEE Security Manager). Without this capability the alternatives are: * Check if all attributes are invisible and if so, hide the directory. Beyond trouble getting this to work [1], this is an ABI change for scenarios if userspace happens to depend on group visibility absent any attributes. I.e. this new capability avoids regression since it does not retroactively apply to existing cases. * Publish an empty /sys/bus/pci/devices/$pdev/tsm/ directory for all PCI devices (i.e. for the case when TSM platform support is present, but device support is absent). Unfortunate that this will be a vestigial empty directory in the vast majority of cases. * Reintroduce usage of runtime calls to sysfs_{create,remove}_group() in the PCI core. Bjorn has already indicated that he does not want to see any growth of pci_sysfs_init() [2]. * Drop the named group and simulate a directory by prefixing all TSM-related attributes with "tsm_". Unfortunate to not use the naming capability of a sysfs group as intended. In comparison, there is a small potential for regression if for some reason an @is_visible() callback had dependencies on how many times it was called. Additionally, it is no longer an error to update a group that does not have its directory already present, and it is no longer a WARN() to remove a group that was never visible. Link: https://lore.kernel.org/all/2024012321-envious-procedure-4a58@gregkh/ [1] Link: https://lore.kernel.org/linux-pci/20231019200110.GA1410324@bhelgaas/ [2] Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

…t of groups The sysfs logic already creates a list of groups for the device, so add the sdw_slave_dev_attr_group group to that list instead of having to do a two-step process of adding a group list and then an individual group. This is a step on the way to moving all of the sysfs attribute handling into the default driver core attribute group logic so that the soundwire core does not have to do any of it manually. Cc: Vinod Koul <vkoul@kernel.org> Cc: Bard Liao <yung-chuan.liao@linux.intel.com> Cc: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Cc: Sanyog Kale <sanyog.r.kale@intel.com> Cc: alsa-devel@alsa-project.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

…utes There's no need to special-case the dp0 sysfs attributes, the is_visible() callback in the attribute group can handle that for us, so add that and add it to the attribute group list making the logic simpler overall. This is a step on the way to moving all of the sysfs attribute handling into the default driver core attribute group logic so that the soundwire core does not have to do any of it manually. Cc: Vinod Koul <vkoul@kernel.org> Cc: Bard Liao <yung-chuan.liao@linux.intel.com> Cc: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Cc: Sanyog Kale <sanyog.r.kale@intel.com> Cc: alsa-devel@alsa-project.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

…ice groups The driver core supports the ability to handle the creation and removal of device-specific sysfs files in a race-free manner. Take advantage of that by converting this driver to use this by moving the sysfs attributes into a group and assigning the dev_groups pointer to it. Cc: Vinod Koul <vkoul@kernel.org> Cc: Bard Liao <yung-chuan.liao@linux.intel.com> Cc: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Cc: Sanyog Kale <sanyog.r.kale@intel.com> Cc: alsa-devel@alsa-project.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Now that sdw_slave_sysfs_init() only calls sdw_slave_sysfs_dpn_init(), just do that instead and remove sdw_slave_sysfs_init() to get it out of the way to save a bit of logic and code size. Cc: Vinod Koul <vkoul@kernel.org> Cc: Bard Liao <yung-chuan.liao@linux.intel.com> Cc: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Cc: Sanyog Kale <sanyog.r.kale@intel.com> Cc: alsa-devel@alsa-project.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Now that we manually created our own attribute group list, the outdated ATTRIBUTE_GROUPS() comments can be removed as they are not needed at all. Cc: Vinod Koul <vkoul@kernel.org> Cc: Bard Liao <yung-chuan.liao@linux.intel.com> Cc: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Cc: Sanyog Kale <sanyog.r.kale@intel.com> Cc: alsa-devel@alsa-project.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

plbossart · 2024-01-31T12:57:17Z

The first patch in this series seems to prevent LunarLake devices from booting to SSH, even with the sof-pci-lnl driver blacklisted. IOW something non-audio is going boink. We would need to test with a direct device access, testing remotely is impossible.

djbw · 2024-02-16T18:08:45Z

@plbossart any update on the debug?

I posted a debug patch to get more info here

plbossart · 2024-02-16T18:39:51Z

@plbossart any update on the debug?

I posted a debug patch to get more info here

Sorry, we've been busy fixing bugs while most of our team was on CNY vacation. @marc-hb is looking into this today.

https://lore.kernel.org/all/65baa2c754e8d_37ad2941f@dwillia2-xfh.jf.intel.com.notmuch/ Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

plbossart · 2024-02-16T22:30:38Z

@djbw the results are the same, the boot process is stopped while dealing with the initramfs. I'll have to leave you in the capable hands of JF-based @marc-hb, it's likely that LNL has some sort of sysfs tree snafu that is exposed by your patch.

djbw · 2024-02-16T22:45:43Z

@djbw the results are the same, the boot process is stopped while dealing with the initramfs. I'll have to leave you in the capable hands of JF-based @marc-hb, it's likely that LNL has some sort of sysfs tree snafu that is exposed by your patch.

Sounds good, I was not expecting the patch to help, but hopefully give more info of which is_visible() callback might have been problematic, but if you don't even get a console / log out of the machine then not much to work with.

marc-hb · 2024-02-17T00:30:03Z

Dan's commit "sysfs: Introduce a mechanism to hide static attribute_groups" is definitely the trigger. In addition to comparing the base versus first commit of this PR, I also cherry picked it over v6.8-rc4 with the very same results: v6.8-rc4 alone works fine; hard hang and no logs at all with the cherry-pick on top.

What this patch triggers is a super hard hang while or immediately after loading the initramfs. So there is absolutely ZERO kernel logs on the eDP panel.

I tried the following kernel params: debug break=top. They work fine without the trigger, the system stops at the start of the initramfs. With the commit, the system crashes too early for these to have any effect.

Without Dan's commit, Ctrl-alt-del and SysRq work fine after break=top. With it, these key strokes have no effect while the kernel is locked up. Using the reset button on the front panel is required.

I changed the initramfs compression in the odd chance. No difference.

https://lore.kernel.org/all/65baa2c754e8d_37ad2941f@dwillia2-xfh.jf.intel.com.notmuch/raw didn't apply, it seems to have spaces instead of tabs.
Pierre fixed that patch for me and shared it in this PR. I added it but it made not difference. I never tried the other, later commits in this PR.

For complete disclosure these ACPI error show up every time the system boots:

Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.BGNV], AE_ALREADY_EXISTS (20230331/dswload2-326)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20230331/psobject-220)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.BGMA], AE_ALREADY_EXISTS (20230331/dsfield-637)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.BGMS], AE_ALREADY_EXISTS (20230331/dsfield-637)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.BGIA], AE_ALREADY_EXISTS (20230331/dsfield-637)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.PC00.PTMA], AE_ALREADY_EXISTS (20230331/dswload2-326)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20230331/psobject-220)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.PC00.PTMS], AE_ALREADY_EXISTS (20230331/dswload2-326)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20230331/psobject-220)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI BIOS Error (bug): Failure creating named object [\_SB.PC00.PTIA], AE_ALREADY_EXISTS (20230331/dswload2-326)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI Error: AE_ALREADY_EXISTS, During name lookup/catalog (20230331/psobject-220)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI Error: Cannot assign type [Device] to [Integer] (must be type Int/Str/Buf) (20230331/exstoren-90)
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI: \: Can't tag data node
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI: \: Can't tag data node
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI: \: Can't tag data node
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI: \: Can't tag data node
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: ACPI: \: Can't tag data node
Feb 17 00:00:46 jf-lnlm-rvp-nocodec-1 kernel: i8042: Can't read CTR while initializing i8042
Feb 17 00:00:50 jf-lnlm-rvp-nocodec-1 kernel: ucsi_acpi USBC000:00: failed to reset PPM!
Feb 17 00:00:50 jf-lnlm-rvp-nocodec-1 kernel: ucsi_acpi USBC000:00: error -ETIMEDOUT: PPM init failed

The next steps could/should be:

Split Dan's commit into even smaller parts and try to isolate which part makes a difference
Find a serial port and try earlyprintk

marc-hb · 2024-02-17T00:49:24Z

For people not familiar with SOF, these test results show no undesirable effect on generations other than LNL:
https://sof-ci.01.org/linuxpr/PR4799/build1464/devicetest/index.html
https://sof-ci.01.org/linuxpr/PR4799/build1465/devicetest/index.html
https://sof-ci.01.org/linuxpr/PR4799/build1466/devicetest/index.html

marc-hb · 2024-02-22T02:16:33Z

I found how to use a serial connection to this board. Crashes in perf_event_sysfs_init every time:

.
.
.
[    5.747097] pci 0000:00:15.1: Adding to iommu group 13
[    5.752857] pci 0000:00:15.2: Adding to iommu group 13
[    5.758619] pci 0000:00:15.3: Adding to iommu group 13
[    5.764395] pci 0000:00:16.0: Adding to iommu group 14
[    5.770246] pci 0000:00:1c.0: Adding to iommu group 15
[    5.776040] pci 0000:00:1f.0: Adding to iommu group 16
[    5.781803] pci 0000:00:1f.3: Adding to iommu group 16
[    5.787568] pci 0000:00:1f.4: Adding to iommu group 16
[    5.793325] pci 0000:00:1f.5: Adding to iommu group 16
[    5.799186] pci 0000:7f:00.0: Adding to iommu group 17
[    5.805494] DMAR: Intel(R) Virtualization Technology for Direg for IO (SWIOTLB)
[    5.819731] software IO TLB: mapped [mem [    5.828831] clocksource: tsc: mask: 0xffffffffffffffff max_cycles: 0x28680fa287f, max_idle_ns: 440795281151 ns
[    5.839895] clocksource: Switched to clocksource tsc
[    5.845484] platform rtc_cmos: registered platform RTC device[    5.857089] BUG: kernel NULL pointer dereference, address: 00rnel mode
[    5.857658] #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI
[    5.857658] CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20
[    5.857658] Code: 00 00 0 5.857658] RSP: 0000:ffffaa1d400f7dd0 EFLAGS: 00010206
[    5.8 R08: fffffffffff742c3 R09: ffff9daec26c0000
[    5.857658] R10ffff9db22ec00000(0000) knlGS:0000000000000000
[    5.857658] CS2: 0000000000000000
[    5.857658] DR3: 0000000000000000 DR6: 0x181/0x490
[    5.857658]  ? exc_page_fault+0x68/0x190
[    5.5.857658]  internal_create_groups+0x42/0xa0
[    5.857658]  pmu_dev_alloc+0xc0/0xe0
[    5.857658]  perf_event_sysfs_init+0x580000000000 ]---
[    5.857658] RIP: 0010:exra_is_visible+0x14/0 90 90 90 90 90 90 90 90 90 f3 0f 1e fa 0f 1f 44 00 00 31 c0 83   5.857658] RBP: 0000000000000001 R08: fffffffffff742c3 R09: fff000000001 R12: ffff9da
ec355f800
[    5.857658] R13: ffffffffab80000000080050033
[    5.857658] CR2: 0000000000000008 CR3: 0000000400
[    5.857658] PKRU: 55555554
[    5.857658] note: swap[    6.235716] Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000009
[    6.236701] Kernel Offset: 0x28c00000

djbw · 2024-02-22T04:23:24Z

I think this is the fixup, I did not expect empty attribute arrays, but that seems to be standard practice for perf pmu.

diff --git a/fs/sysfs/group.c b/fs/sysfs/group.c
index ccb275cdabcb..8c63ba3cfc47 100644
--- a/fs/sysfs/group.c
+++ b/fs/sysfs/group.c
@@ -33,10 +33,10 @@ static void remove_files(struct kernfs_node *parent,
 
 static umode_t __first_visible(const struct attribute_group *grp, struct kobject *kobj)
 {
-	if (grp->attrs && grp->is_visible)
+	if (grp->attrs && grp->attrs[0] && grp->is_visible)
 		return grp->is_visible(kobj, grp->attrs[0], 0);
 
-	if (grp->bin_attrs && grp->is_bin_visible)
+	if (grp->bin_attrs && grp->bin_attrs[0] && grp->is_bin_visible)
 		return grp->is_bin_visible(kobj, grp->bin_attrs[0], 0);
 
 	return 0;
-- 
2.43.0

As no one ever wanted to boot from a microphone, it's very rare that we have to use the serial console to debug very early boot issues. But it can happen, see for instance thesofproject/linux#4799 In such case it's very convenient to use another test device to collect logs on the serial console. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: "Rafael J. Wysocki" <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com>

marc-hb · 2024-02-22T22:53:34Z

I think this is the fixup, I did not expect empty attribute arrays, but that seems to be standard practice for perf pmu.

This fixup works. System boots and nothing suspicious in the logs.

As no one ever wanted to boot from a microphone, it's very rare that we have to use the serial console to debug very early boot issues. But it can happen, see for instance thesofproject/linux#4799 In such case it's very convenient to use another test device to collect logs on the serial console. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

plbossart · 2024-02-23T00:40:48Z

closing since there are no issues remaining

It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

JIRA: https://issues.redhat.com/browse/RHEL-102859 commit cd69fed Author: Dan Williams <dan.j.williams@intel.com> Date: Thu Feb 22 12:40:54 2024 -0800 sysfs: Fix crash on empty group attributes array It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: David Arcari <darcari@redhat.com>

commit cd69fed upstream. It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Intel-SIG: commit cd69fed sysfs: Fix crash on empty group attributes array Backport PMT spec 3.0 support (Discovery driver and preparation for PMT AET) Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> [ Zhang Rui: amend commit log ] Signed-off-by: Zhang Rui <rui.zhang@intel.com>

mainline inclusion from mainline-v6.9-rc1 commit cd69fedf58f8ab1ab511f7c6ac1969cebf1c935f category: feature bugzilla: https://atomgit.com/openeuler/intel-kernel/issues/186 CVE: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=cd69fedf58f8ab1ab511f7c6ac1969cebf1c935f ------------------------------------------------- Intel-SIG: commit cd69fedf58f8 sysfs: Fix crash on empty group attributes array Backport PMT spec 3.0 support (Discovery driver and preparation for PMT AET) It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd24b41 ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> [ Zhang Rui: amend commit log ] Signed-off-by: Zhang Rui <rui.zhang@intel.com>

commit cd69fed upstream. It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Intel-SIG: commit cd69fed sysfs: Fix crash on empty group attributes array Backport PMT spec 3.0 support (Discovery driver and preparation for PMT AET) Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> [ Zhang Rui: amend commit log ] Signed-off-by: Zhang Rui <rui.zhang@intel.com>

commit cd69fed upstream. It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: PvsNarasimha <PVS.NarasimhaRao@amd.com>

commit cd69fed upstream. It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Guixiong Wei <weiguixiong@bytedance.com>

It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Bug: 492018312 Change-Id: Ie71f5ccbe84589db9b803bc376c129bdba20b1ef (cherry picked from commit cd69fed) Signed-off-by: Selvarasu Ganesan <selvarasu.g@samsung.com>

It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd24b41 ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Bug: 492018312 Change-Id: Ie71f5ccbe84589db9b803bc376c129bdba20b1ef (cherry picked from commit cd69fedf58f8ab1ab511f7c6ac1969cebf1c935f) Signed-off-by: Selvarasu Ganesan <selvarasu.g@samsung.com>

It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Bug: 492018312 Change-Id: Ie71f5ccbe84589db9b803bc376c129bdba20b1ef (cherry picked from commit cd69fed) Signed-off-by: Selvarasu Ganesan <selvarasu.g@samsung.com>

It turns out that arch/x86/events/intel/core.c makes use of "empty" attributes. static struct attribute *empty_attrs; __init int intel_pmu_init(void) { struct attribute **extra_skl_attr = &empty_attrs; struct attribute **extra_attr = &empty_attrs; struct attribute **td_attr = &empty_attrs; struct attribute **mem_attr = &empty_attrs; struct attribute **tsx_attr = &empty_attrs; ... That breaks the assumption __first_visible() that expects that if grp->attrs is set then grp->attrs[0] must also be set and results in backtraces like: BUG: kernel NULL pointer dereference, address: 00rnel mode #PF: error_code(0x0000) - not-present ] PREEMPT SMP NOPTI CPU: 1 PID: 1 Comm: swapper/IP: 0010:exra_is_visible+0x14/0x20 ? exc_page_fault+0x68/0x190 internal_create_groups+0x42/0xa0 pmu_dev_alloc+0xc0/0xe0 perf_event_sysfs_init+0x580000000000 ]--- RIP: 0010:exra_is_visible+0x14/0 Check for non-empty attributes array before calling is_visible(). Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Closes: thesofproject/linux#4799 (comment) Fixes: 70317fd24b41 ("sysfs: Introduce a mechanism to hide static attribute_groups") Cc: Marc Herbert <marc.herbert@intel.com> Cc: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Tested-by: Marc Herbert <marc.herbert@intel.com> Link: https://lore.kernel.org/r/170863445442.1479840.1818801787239831650.stgit@dwillia2-xfh.jf.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Bug: 492018312 Change-Id: Ie71f5ccbe84589db9b803bc376c129bdba20b1ef (cherry picked from commit cd69fedf58f8ab1ab511f7c6ac1969cebf1c935f) Signed-off-by: Selvarasu Ganesan <selvarasu.g@samsung.com>

djbw and others added 6 commits January 31, 2024 09:05

sysfs: add debug patch suggested by Dan Williams

8960d1a

https://lore.kernel.org/all/65baa2c754e8d_37ad2941f@dwillia2-xfh.jf.intel.com.notmuch/ Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

marc-hb mentioned this pull request Feb 22, 2024

sof-dev-config: add CONFIG_USB_SERIAL_CP210X=m thesofproject/kconfig#90

Merged

plbossart closed this Feb 23, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[TEST] SoundWire sysfs upstream changes#4799

[TEST] SoundWire sysfs upstream changes#4799
plbossart wants to merge 7 commits intothesofproject:topic/sof-devfrom
plbossart:test/sysfs-upstream

plbossart commented Jan 31, 2024

Uh oh!

plbossart commented Jan 31, 2024

Uh oh!

djbw commented Feb 16, 2024

Uh oh!

plbossart commented Feb 16, 2024

Uh oh!

plbossart commented Feb 16, 2024

Uh oh!

djbw commented Feb 16, 2024

Uh oh!

marc-hb commented Feb 17, 2024

Uh oh!

marc-hb commented Feb 17, 2024 •

edited

Loading

Uh oh!

marc-hb commented Feb 22, 2024

Uh oh!

djbw commented Feb 22, 2024 •

edited

Loading

Uh oh!

marc-hb commented Feb 22, 2024 •

edited

Loading

Uh oh!

plbossart commented Feb 23, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

plbossart commented Jan 31, 2024

Uh oh!

plbossart commented Jan 31, 2024

Uh oh!

djbw commented Feb 16, 2024

Uh oh!

plbossart commented Feb 16, 2024

Uh oh!

plbossart commented Feb 16, 2024

Uh oh!

djbw commented Feb 16, 2024

Uh oh!

marc-hb commented Feb 17, 2024

Uh oh!

marc-hb commented Feb 17, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

marc-hb commented Feb 22, 2024

Uh oh!

djbw commented Feb 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

marc-hb commented Feb 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

plbossart commented Feb 23, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

marc-hb commented Feb 17, 2024 •

edited

Loading

djbw commented Feb 22, 2024 •

edited

Loading

marc-hb commented Feb 22, 2024 •

edited

Loading