| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <aVn9K96d5CD5m_Y4@venus>
Date: Sun, 4 Jan 2026 07:01:26 +0100
From: Sebastian Reichel <sebastian.reichel@...labora.com>
To: Shawn Lin <shawn.lin@...k-chips.com>
Cc: "ping.gao" <ping.gao@...sung.com>, jh80.chung@...sung.com,
ulf.hansson@...aro.org, linux-kernel@...r.kernel.org, linux-mmc@...r.kernel.org
Subject: Re: [PATCH] MMC:dw_mmc:when enable biu_clk should check whether this
clk is normal
Hi,
On Sun, Jan 04, 2026 at 10:55:48AM +0800, Shawn Lin wrote:
> 在 2026/01/04 星期日 10:38, ping.gao 写道:
> > some vendor doesn't use biu_clk , enable it will panic
> > log is below:
>
> I still don't understand the real issue you tried to solve.
>
> host->biu_clk is got via devm_clk_get() in dw_mci_probe(). If
> succeed, immediately it's enabled a few lines below. Otherwise,
> dw_mci_probe() returns with failure. Seems you are not enabling it in
> runtime callbacks, but why clk_prepare_enable(host->biu_clk) is able
> to work when driver probed?
Based on the error and this patch, I believe what happens is:
1. dw_mci_probe() calls devm_clk_get(host->dev, "biu")
2. The clock does not exist on the platform from Ping Gao, so
host->biu_clk gets an error pointer assigned
3. "biu clock not available" gets printed at debug level
4. the error is ignored, but the error pointer is still assigned to
host->biu_clk
5. in the runtime PM handler clk_prepare_enable is called for
host->biu_clk and thus for an error pointer
Instead of hacking this up even further (as this patch does),
the proper fix is to cleanup the probe function. I believe
the complete biu clock handling in it could be reduced to this:
host->biu_clk = devm_clk_get_optional_enabled(host->dev, "biu");
ret = PTR_ERR_OR_ZERO(host->biu_clk);
if (ret)
return dev_err_probe(host->dev, ret, "failed to get biu clock\n");
That will assign NULL to host->biu_clk when there is no clock
defined and it is fine to run clk_prepare_enable() with NULL
as clock pointer. Note, that this will handle the clock enabling
as part of devm, so you will also have to drop the
clk_disable_unprepare() calls from the err_clk_biu goto label
and the extra call from dw_mci_remove().
Something similar should probably be done for host->ciu_clk, which
seems to have the same incorrect logic.
Greetings,
-- Sebastian
> > [ 438.400868] [7: binder:436_2: 4998] Unable to handle kernel paging request at virtual address fffffffffffffffe
> > [ 438.400877] [7: binder:436_2: 4998] Mem abort info:
> > [ 438.400881] [7: binder:436_2: 4998] ESR = 0x0000000096000005
> > [ 438.400887] [7: binder:436_2: 4998] EC = 0x25: DABT (current EL), IL = 32 bits
> > [ 438.400894] [7: binder:436_2: 4998] SET = 0, FnV = 0
> > [ 438.400899] [7: binder:436_2: 4998] EA = 0, S1PTW = 0
> > [ 438.400904] [7: binder:436_2: 4998] FSC = 0x05: level 1 translation fault
> > ...
> > [ 438.409424] [7: binder:436_2: 4998] Call trace:
> > [ 438.409429] [7: binder:436_2: 4998] clk_prepare+0x10/0x24
> > [ 438.409439] [7: binder:436_2: 4998] dw_mci_runtime_resume+0x50/0x2d8 [dw_mmc_samsung cd210e210975263404c28fc89778f369f8398f0c]
> > [ 438.409471] [7: binder:436_2: 4998] dw_mci_exynos_runtime_resume+0x18/0x58 [dw_mmc_exynos_samsung 2735a594c7c9c9e8c65b0b87523fbf70dcaabfff]
> > [ 438.409496] [7: binder:436_2: 4998] pm_generic_runtime_resume+0x40/0x58
> > [ 438.409506] [7: binder:436_2: 4998] pm_runtime_force_resume+0x9c/0x134
> > [ 438.409517] [7: binder:436_2: 4998] platform_pm_resume+0x40/0x8c
> > [ 438.409529] [7: binder:436_2: 4998] dpm_run_callback+0x64/0x230
> > [ 438.409540] [7: binder:436_2: 4998] __device_resume+0x1d8/0x394
> > [ 438.409551] [7: binder:436_2: 4998] dpm_resume+0x110/0x2b8
> > [ 438.409561] [7: binder:436_2: 4998] dpm_resume_end+0x1c/0x38
> > [ 438.409570] [7: binder:436_2: 4998] suspend_devices_and_enter+0x828/0xab0
> > [ 438.409582] [7: binder:436_2: 4998] pm_suspend+0x334/0x618
> > [ 438.409592] [7: binder:436_2: 4998] state_store+0x104/0x144
> > [ 438.409601] [7: binder:436_2: 4998] kobj_attr_store+0x30/0x48
> > [ 438.409610] [7: binder:436_2: 4998] sysfs_kf_write+0x54/0x6c
> > [ 438.409619] [7: binder:436_2: 4998] kernfs_fop_write_iter+0x104/0x1a8
> > [ 438.409628] [7: binder:436_2: 4998] vfs_write+0x24c/0x2f4
> > [ 438.409640] [7: binder:436_2: 4998] ksys_write+0x78/0xe8
> > [ 438.409652] [7: binder:436_2: 4998] __arm64_sys_write+0x1c/0x2c
> > [ 438.409664] [7: binder:436_2: 4998] invoke_syscall+0x58/0x114
> > [ 438.409676] [7: binder:436_2: 4998] el0_svc_common+0xac/0xe0
> > [ 438.409687] [7: binder:436_2: 4998] do_el0_svc+0x1c/0x28
> > [ 438.409698] [7: binder:436_2: 4998] el0_svc+0x38/0x68
> > [ 438.409705] [7: binder:436_2: 4998] el0t_64_sync_handler+0x68/0xbc
> > [ 438.409712] [7: binder:436_2: 4998] el0t_64_sync+0x1a8/0x1ac
> > Signed-off-by: ping.gao <ping.gao@...sung.com>
> > ---
> > drivers/mmc/host/dw_mmc.c | 12 +++++++++---
> > 1 file changed, 9 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/mmc/host/dw_mmc.c b/drivers/mmc/host/dw_mmc.c
> > index c5db92bbb094..61f6986f15ef 100644
> > --- a/drivers/mmc/host/dw_mmc.c
> > +++ b/drivers/mmc/host/dw_mmc.c
> > @@ -3633,9 +3633,15 @@ int dw_mci_runtime_resume(struct device *dev)
> > if (host->slot &&
> > (mmc_host_can_gpio_cd(host->slot->mmc) ||
> > !mmc_card_is_removable(host->slot->mmc))) {
> > - ret = clk_prepare_enable(host->biu_clk);
> > - if (ret)
> > - return ret;
> > + if (IS_ERR(host->biu_clk)) {
> > + dev_dbg(host->dev, "biu clock not available\n");
> > + } else {
> > + ret = clk_prepare_enable(host->biu_clk);
> > + if (ret) {
> > + dev_err(host->dev, "failed to enable biu clock\n");
> > + goto err;
> > + }
> > + }
> > }
> > ret = clk_prepare_enable(host->ciu_clk);
>
>
Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)
Powered by blists - more mailing lists