Re: [PATCH] usb: typec: add sanity check for opmode

[lijiazi <jqqlijiazi@gmail.com>]

On Thu, Nov 26, 2020 at 11:28:57AM +0100, Greg Kroah-Hartman wrote:
> On Thu, Nov 26, 2020 at 06:00:33PM +0800, lijiazi wrote:
> > On Thu, Nov 26, 2020 at 09:24:29AM +0100, Greg Kroah-Hartman wrote:
> > > On Thu, Nov 26, 2020 at 01:10:55PM +0800, lijiazi wrote:
> > > > If usb drivers set a invalid value, for example, a negative
> > > > value. then a userspace task
> > > > cat sys/class/typec/port0/power_operation_mode, will cause a
> > > > panic issue:
> > > > [154325.262827] Unable to handle kernel paging request at virtual
> > > > address ffffff980aad8b68
> > > > [154325.262838] Mem abort info:
> > > > [154325.262843]   ESR = 0x96000005
> > > > [154325.262849]   Exception class = DABT (current EL), IL = 32 bits
> > > > [154325.262855]   SET = 0, FnV = 0
> > > > [154325.262860]   EA = 0, S1PTW = 0
> > > > [154325.262865] Data abort info:
> > > > [154325.262870]   ISV = 0, ISS = 0x00000005
> > > > [154325.262875]   CM = 0, WnR = 0
> > > > [154325.262880] swapper pgtable: 4k pages, 39-bit VAs, pgdp =
> > > > 000000001cae9c14
> > > > [154325.262884] [ffffff980aad8b68] pgd=0000000000000000,
> > > > pud=0000000000000000
> > > > [154325.262891] Internal error: Oops: 96000005 [#1] PREEMPT SMP
> > > > [154325.262896] Modules linked in: rmnet_perf(O) rmnet_shs(O)
> > > > wlan(O) sla(O) exfat(O) machine_dlkm(O) tfa98xx_dlkm(O)
> > > > cs35l41_dlkm(O) wcd938x_slave_dlkm(O) wcd938x_dlkm(O)
> > > > wcd9xxx_dlkm(O) mbhc_dlkm(O) tx_macro_dlkm(O) rx_macro_dlkm(O)
> > > > va_macro_dlkm(O) wsa_macro_dlkm(O) swr_ctrl_dlkm(O)
> > > > bolero_cdc_dlkm(O) wsa881x_dlkm(O) wcd_core_dlkm(O) stub_dlkm(O)
> > > > hdmi_dlkm(O) swr_dlkm(O) pinctrl_lpi_dlkm(O) pinctrl_wcd_dlkm(O)
> > > > usf_dlkm(O) native_dlkm(O) platform_dlkm(O) q6_dlkm(O)
> > > > adsp_loader_dlkm(O) apr_dlkm(O) snd_event_dlkm(O)
> > > > q6_notifier_dlkm(O) q6_pdr_dlkm(O) [last unloaded: rmnet_perf]
> > > > [154325.262939] Process usb@1.2-service (pid: 2501, stack limit
> > > > 		= 0x00000000cb0343ac)
> > > > [154325.262946] CPU: 5 PID: 2501 Comm: usb@1.2-service Tainted:
> > > > G S      W  O      4.19.113-perf-g0307705d321bc #1
> > > > [154325.262950] Hardware name: Qualcomm Technologies, Inc.
> > > > xiaomi cas (DT)
> > > > [154325.262955] pstate: 80400005 (Nzcv daif +PAN -UAO)
> > > > [154325.262967] pc : power_operation_mode_show+0x34/0x58
> > > > [154325.262971] lr : power_operation_mode_show+0x34/0x58
> > > > [154325.262974] sp : ffffffc1376bf990
> > > > [154325.262977] x29: ffffffc1376bf990 x28: ffffffc195aede58
> > > > [154325.262981] x27: ffffffc195aede48 x26: ffffff9008c55638
> > > > [154325.262986] x25: ffffffc05a44b300 x24: 0000000000001000
> > > > [154325.262990] x23: ffffffc19a340018 x22: ffffffc19a340018
> > > > [154325.262994] x21: ffffff900bb10038 x20: ffffff980aad8b68
> > > > [154325.262998] x19: ffffffc05a44b300 x18: 0000000000000000
> > > > [154325.263002] x17: 0000000000000000 x16: 0000000000000000
> > > > [154325.263006] x15: 0000000000000000 x14: 00000000080a38f8
> > > > [154325.263010] x13: ffffff880b489860 x12: 0000000000000000
> > > > [154325.263014] x11: 0000000000000000 x10: 1ffffff30155b16d
> > > > [154325.263018] x9 : 0000000000000000 x8 : 0000000000000007
> > > > [154325.263022] x7 : 0000000000000000 x6 : 000000000000003f
> > > > [154325.263026] x5 : 0000000000000040 x4 : 0000000000000000
> > > > [154325.263030] x3 : 0000000000000004 x2 : ffffffc05a44b300
> > > > [154325.263034] x1 : ffffff900bb10038 x0 : ffffff980aad8b68
> > > > [154325.263039]
> > > > [154325.263039] SP: 0xffffffc1376bf910:
> > > > [154325.263043] f910  00001000 00000000 5a44b300 ffffffc0
> > > > 08c55638 ffffff90 95aede48 ffffffc1
> > > > [154325.263052] f930  95aede58 ffffffc1 376bf990 ffffffc1
> > > > 09024b5c ffffff90 376bf990 ffffffc1
> > > > [154325.263060] f950  09024b5c ffffff90 80400005 00000000
> > > > 376bf9c0 ffffffc1 083cdf64 ffffff90
> > > > [154325.263067] f970  ffffffff 0000007f 376bfd80 ffffffc1
> > > > 376bf990 ffffffc1 09024b5c ffffff90
> > > > [154325.263075] f990  376bf9b0 ffffffc1 08c55678 ffffff90
> > > > 0bb10048 ffffff90 5a44b300 ffffffc0
> > > > [154325.263082] f9b0  376bf9e0 ffffffc1 085075e0 ffffff90
> > > > 0a5d43f8 ffffff90 9a9d9f00 ffffffc1
> > > > [154325.263090] f9d0  95aede48 ffffffc1 95aede58 ffffffc1
> > > > 376bfa30 ffffffc1 08506248 ffffff90
> > > > [154325.263097] f9f0  95aede40 ffffffc1 00000000 00000000
> > > > 00000001 00000000 9a9d9f00 ffffffc1
> > > > [154325.263106] Call trace:
> > > > [154325.263111]  power_operation_mode_show+0x34/0x58
> > > > [154325.263116]  dev_attr_show+0x40/0x80
> > > > [154325.263123]  sysfs_kf_seq_show+0x110/0x1c0
> > > > [154325.263127]  kernfs_seq_show+0x80/0x98
> > > > [154325.263133]  seq_read+0x2d8/0x778
> > > > [154325.263136]  kernfs_fop_read+0xa4/0x2a8
> > > > [154325.263142]  __vfs_read+0xd4/0x2e8
> > > > [154325.263145]  vfs_read+0xe0/0x1b0
> > > > [154325.263149]  ksys_read+0xdc/0x170
> > > > [154325.263153]  __arm64_sys_read+0x44/0x58
> > > > [154325.263158]  el0_svc_common+0xd8/0x1d0
> > > > [154325.263162]  el0_svc_handler+0x90/0xb8
> > > > [154325.263166]  el0_svc+0x8/0xc
> > > > 
> > > > Signed-off-by: lijiazi <lijiazi@xiaomi.com>
> > > 
> > > I need a name here that you sign legal documents with.  If this is
> > > "ligiazi", that's fine, but at least capitalize it?
> > >
> > 
> > My Name is Jiazi Li.
> 
> Great, please fix up your "From:" line, and this signed-off-by line to
> have that name there, as is documented.
> 
OK, I will fix up "From:" and "Signed-off-by" line.
Thanks!
> > > > ---
> > > >  drivers/usb/typec/class.c | 2 ++
> > > >  include/linux/usb/typec.h | 1 +
> > > >  2 files changed, 3 insertions(+)
> > > > 
> > > > diff --git a/drivers/usb/typec/class.c b/drivers/usb/typec/class.c
> > > > index cb1362187..bcc8f79 100644
> > > > --- a/drivers/usb/typec/class.c
> > > > +++ b/drivers/usb/typec/class.c
> > > > @@ -1564,6 +1564,8 @@ void typec_set_pwr_opmode(struct typec_port *port,
> > > >  {
> > > >  	struct device *partner_dev;
> > > >  
> > > > +	if (opmode >= TYPEC_PWR_MODE_MAX)
> > > > +		return;
> > > 
> > > Shouldn't we report the error somehow to userspace so they know they
> > > have a broken device that needs to be fixed?
> > >
> > 
> > This typec_set_pwr_opmode function has no return value, so just return.
> 
> But should that change?  Shouldn't the operation fail somehow?
> 
Yes, if opmode invalid returns -EINVAL, 0 is returned in other cases.
What do you think about it?
> > > >  	if (port->pwr_opmode == opmode)
> > > >  		return;
> > > >  
> > > > diff --git a/include/linux/usb/typec.h b/include/linux/usb/typec.h
> > > > index 5447532..e21c791 100644
> > > > --- a/include/linux/usb/typec.h
> > > > +++ b/include/linux/usb/typec.h
> > > > @@ -56,6 +56,7 @@ enum typec_pwr_opmode {
> > > >  	TYPEC_PWR_MODE_1_5A,
> > > >  	TYPEC_PWR_MODE_3_0A,
> > > >  	TYPEC_PWR_MODE_PD,
> > > > +	TYPEC_PWR_MODE_MAX,
> > > 
> > > If these are specific values that a device uses, then please set them to
> > > the specific values and don't just rely on the compiler to set the
> > > correctly.
> > >
> > 
> > This invalid value is generated by a faulty usb driver logic, so it
> > could be any vaule.  
> 
> Yes, but the valid values are specific numbers, those should be
> specified here.
> 
There is not new mode value, in my device just use these four modes.
If has new valid value, I will specified here.
Thanks.
> thanks,
> 
> greg k-h