Hi Dan,
On Wed, 6 Aug 2025 at 20:24, Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:
>
> On Tue, Aug 05, 2025 at 12:50:28AM +0530, Naresh Kamboju wrote:
> > While booting and testing selftest cgroups and filesystem testing on arm64
> > dragonboard-410c the following kernel warnings / errors noticed and system
> > halted and did not recover with selftests Kconfig enabled running the kernel
> > Linux next tag next-20250804.
> >
> > Regression Analysis:
> > - New regression? Yes
> > - Reproducibility? Re-validation is in progress
> >
> > First seen on the next-20250804
> > Good: next-20250801
> > Bad: next-20250804
> >
> > Test regression: next-20250804 Unable to handle kernel execute from
> > non-executable memory at virtual address idem_hash
> > Test regression: next-20250804 refcount_t: addition on 0;
> > use-after-free refcount_warn_saturate
> >
> > Reported-by: Linux Kernel Functional Testing <lkft@xxxxxxxxxx>
> >
> > ## Test crash log
> > [ 9.811341] Unable to handle kernel NULL pointer dereference at
> > virtual address 000000000000002e
> > [ 9.811444] Mem abort info:
> > [ 9.821150] ESR = 0x0000000096000004
> > [ 9.833499] SET = 0, FnV = 0
> > [ 9.833566] EA = 0, S1PTW = 0
> > [ 9.835511] FSC = 0x04: level 0 translation fault
> > [ 9.838901] Data abort info:
> > [ 9.843788] ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
> > [ 9.846565] CM = 0, WnR = 0, TnD = 0, TagAccess = 0
> > [ 9.851938] GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
> > [ 9.853510] rtc-pm8xxx 200f000.spmi:pmic@0:rtc@6000: registered as rtc0
> > [ 9.856992] user pgtable: 4k pages, 48-bit VAs, pgdp=00000000856f8000
> > [ 9.862446] rtc-pm8xxx 200f000.spmi:pmic@0:rtc@6000: setting system
> > clock to 1970-01-01T00:00:31 UTC (31)
> > [ 9.868789] [000000000000002e] pgd=0000000000000000, p4d=0000000000000000
> > [ 9.875459] Internal error: Oops: 0000000096000004 [#1] SMP
> > [ 9.889547] input: pm8941_pwrkey as
> > /devices/platform/soc@0/200f000.spmi/spmi-0/0-00/200f000.spmi:pmic@0:pon@800/200f000.spmi:pmic@0:pon@800:pwrkey/input/input1
> > [ 9.891545] Modules linked in: qcom_spmi_temp_alarm rtc_pm8xxx
> > qcom_pon(+) qcom_pil_info videobuf2_dma_sg ubwc_config qcom_q6v5
> > venus_core(+) qcom_sysmon qcom_spmi_vadc v4l2_fwnode llcc_qcom
> > v4l2_async qcom_vadc_common qcom_common ocmem v4l2_mem2mem drm_gpuvm
> > videobuf2_memops qcom_glink_smem videobuf2_v4l2 drm_exec mdt_loader
> > qmi_helpers gpu_sched drm_dp_aux_bus qnoc_msm8916 videodev
> > drm_display_helper qcom_stats videobuf2_common cec qcom_rng
> > drm_client_lib mc phy_qcom_usb_hs socinfo rpmsg_ctrl display_connector
> > rpmsg_char ramoops rmtfs_mem reed_solomon drm_kms_helper fuse drm
> > backlight
> > [ 9.912286] input: pm8941_resin as
> > /devices/platform/soc@0/200f000.spmi/spmi-0/0-00/200f000.spmi:pmic@0:pon@800/200f000.spmi:pmic@0:pon@800:resin/input/input2
> > [ 9.941186] CPU: 2 UID: 0 PID: 221 Comm: (udev-worker) Not tainted
> > 6.16.0-next-20250804 #1 PREEMPT
> > [ 9.941200] Hardware name: Qualcomm Technologies, Inc. APQ 8016 SBC (DT)
> > [ 9.941206] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> > [ 9.941215] pc : dev_pm_opp_put (/builds/linux/drivers/opp/core.c:1685)
> > [ 9.941233] lr : core_clks_enable+0x54/0x148 venus_core
> > [ 10.004266] sp : ffff8000842b35f0
> > [ 10.004273] x29: ffff8000842b35f0 x28: ffff8000842b3ba0 x27: ffff0000047be938
> > [ 10.004289] x26: 0000000000000000 x25: 0000000000000000 x24: ffff80007b350ba0
> > [ 10.004303] x23: ffff00000ba380c8 x22: ffff00000ba38080 x21: 0000000000000000
> > [ 10.004316] x20: 0000000000000000 x19: ffffffffffffffee x18: 00000000ffffffff
> > [ 10.004330] x17: 0000000000000000 x16: 1fffe000017541a1 x15: ffff8000842b3560
> > [ 10.004344] x14: 0000000000000000 x13: 007473696c5f7974 x12: 696e696666615f65
> > [ 10.004358] x11: 00000000000000c0 x10: 0000000000000020 x9 : ffff80007b33f2bc
> > [ 10.004371] x8 : ffffffffffffffde x7 : ffff0000044a4800 x6 : 0000000000000000
> > [ 10.004384] x5 : 0000000000000002 x4 : 00000000c0000000 x3 : 0000000000000001
> > [ 10.004397] x2 : 0000000000000002 x1 : ffffffffffffffde x0 : ffffffffffffffee
> > [ 10.004412] Call trace:
> > [ 10.004417] dev_pm_opp_put (/builds/linux/drivers/opp/core.c:1685) (P)
> > [ 10.004435] core_clks_enable+0x54/0x148 venus_core
> > [ 10.004504] core_power_v1+0x78/0x90 venus_core
> > [ 10.004560] venus_runtime_resume+0x6c/0x98 venus_core
> > [ 10.004616] pm_generic_runtime_resume
>
> Could you try adding some error checking to core_clks_enable()?
> Does the patch below help?
Your patch works.
The attached patch from Sasha fixes this reported problem on today's
Linux next tag.
$ git log --oneline next-20250805..next-20250807 --
drivers/media/platform/qcom/venus/pm_helpers.c
7881cd6886a89 media: venus: Fix OPP table error handling
- Naresh
commit 7881cd6886a89eda848192d3f5759ce08672e084
Author: Sasha Levin <sashal@xxxxxxxxxx>
Date: Tue Aug 5 08:58:20 2025 -0400
media: venus: Fix OPP table error handling
The venus driver fails to check if dev_pm_opp_find_freq_{ceil,floor}()
returns an error pointer before calling dev_pm_opp_put(). This causes
a crash when OPP tables are not present in device tree.
Unable to handle kernel access to user memory outside uaccess routines
at virtual address 000000000000002e
...
pc : dev_pm_opp_put+0x1c/0x4c
lr : core_clks_enable+0x4c/0x16c [venus_core]
Add IS_ERR() checks before calling dev_pm_opp_put() to avoid
dereferencing error pointers.
Fixes: b179234b5e59 ("media: venus: pm_helpers: use opp-table for the frequency")
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
Signed-off-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
diff --git a/drivers/media/platform/qcom/venus/pm_helpers.c b/drivers/media/platform/qcom/venus/pm_helpers.c
index 8dd5a9b0d060c..e32f8862a9f90 100644
--- a/drivers/media/platform/qcom/venus/pm_helpers.c
+++ b/drivers/media/platform/qcom/venus/pm_helpers.c
@@ -48,7 +48,8 @@ static int core_clks_enable(struct venus_core *core)
int ret;
opp = dev_pm_opp_find_freq_ceil(dev, &freq);
- dev_pm_opp_put(opp);
+ if (!IS_ERR(opp))
+ dev_pm_opp_put(opp);
for (i = 0; i < res->clks_num; i++) {
if (IS_V6(core)) {
@@ -660,7 +661,8 @@ static int decide_core(struct venus_inst *inst)
/*TODO : divide this inst->load by work_route */
opp = dev_pm_opp_find_freq_floor(dev, &max_freq);
- dev_pm_opp_put(opp);
+ if (!IS_ERR(opp))
+ dev_pm_opp_put(opp);
min_loaded_core(inst, &min_coreid, &min_load, false);
min_loaded_core(inst, &min_lp_coreid, &min_lp_load, true);
@@ -1121,7 +1123,8 @@ static int load_scale_v4(struct venus_inst *inst)
freq = max(freq_core1, freq_core2);
opp = dev_pm_opp_find_freq_floor(dev, &max_freq);
- dev_pm_opp_put(opp);
+ if (!IS_ERR(opp))
+ dev_pm_opp_put(opp);
if (freq > max_freq) {
dev_dbg(dev, VDBGL "requested clock rate: %lu scaling clock rate : %lu\n",
@@ -1131,7 +1134,8 @@ static int load_scale_v4(struct venus_inst *inst)
}
opp = dev_pm_opp_find_freq_ceil(dev, &freq);
- dev_pm_opp_put(opp);
+ if (!IS_ERR(opp))
+ dev_pm_opp_put(opp);
set_freq:
