linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Bin Liu <b-liu@ti.com>
To: SAURAV GIREPUNJE <saurav.girepunje@gmail.com>
Cc: <gregkh@linuxfoundation.org>, <linux-usb@vger.kernel.org>,
	<linux-kernel@vger.kernel.org>, <saurav.girepunje@hotmail.com>
Subject: Re: [PATCH] usb: musb: Fix external abort on non-linefetch
Date: Thu, 6 Feb 2020 13:17:35 -0600	[thread overview]
Message-ID: <20200206191735.GA11124@iaqt7> (raw)
In-Reply-To: <20200119060404.GA104504@google.com>

Hi Saurav,

On Sun, Jan 19, 2020 at 11:34:04AM +0530, SAURAV GIREPUNJE wrote:
> Hi Bin,
> 
> Yes, I really got this kernel dump for mode_show() and vbus_store(),
> While accessing show/store for mode and vbus through sysfs .

Please provide the kernel dump for mode_show() and vbus_store(). The log
below is for mode_store().

> I have submitted earlier also three patches for same bug, In reply to
> one of those patches you
> asked me to merge another two patches and just sent in one.

I don't have issue to merge all the fixes into one patch, since they fix
the same bug. But I need the dump log to see how the dump is triggered
in mode_show() and vbus_store() as I don't see the how it could happen
and I cannot reproduce it. These two function do not access any musb
register.

By the way, please don't top-posting, and reply to my message to
preserve the context.

-Bin.

> 
> Regards,
> Saurav Girepunje
> 
> On 27/10/19 14:06 +0530, Saurav Girepunje wrote:
> > While setting the usb mode from sysfs. Below error came on kernel
> > version 4.19.
> > 
> > On latest kernel vserion api name changed. Therefore API
> > name and backtrace API name are different.
> > 
> >    [  821.908066] Backtrace:
> >    [  821.910695] [<bf078fc0>] (musb_default_readl [musb_hdrc]) from [<bf0af738>] (dsps_musb_set_mode+0x38/0x12c [musb_dsps])
> >    [  821.922059] [<bf0af700>] (dsps_musb_set_mode [musb_dsps]) from [<bf07899c>] (musb_mode_store+0xc8/0x12c [musb_hdrc])
> >    [  821.933105]  r7:a0010013 r6:0000000b r5:cd79d200 r4:cb634010
> >    [  821.939096] [<bf0788d4>] (musb_mode_store [musb_hdrc]) from [<c0425184>] (dev_attr_store+0x20/0x2c)
> >    [  821.948593]  r7:cd79d200 r6:c5abbf78 r5:00000000 r4:bf0788d4
> >    [  821.954549] [<c0425164>] (dev_attr_store) from [<c0285b08>] (sysfs_kf_write+0x48/0x4c)
> >    [  821.962859]  r5:00000000 r4:c0425164
> >    [  821.966620] [<c0285ac0>] (sysfs_kf_write) from [<c0285274>] (kernfs_fop_write+0xfc/0x1fc)
> >    [  821.975200]  r5:00000000 r4:cd79d080
> >    [  821.978966] [<c0285178>] (kernfs_fop_write) from [<c020ec00>] (__vfs_write+0x34/0x120)
> >    [  821.987280]  r10:00000000 r9:0000000b r8:00000000 r7:0000000b r6:c5abbf78 r5:c0285178
> >    [  821.995493]  r4:cb67a3c0
> >    [  821.998160] [<c020ebcc>] (__vfs_write) from [<c020fae8>] (vfs_write+0xa8/0x170)
> >    [  822.005835]  r9:0000000b r8:00000000 r7:c5abbf78 r6:000def80 r5:cb67a3c0 r4:0000000b
> >    [  822.013969] [<c020fa40>] (vfs_write) from [<c02108d0>] (SyS_write+0x44/0x98)
> >    [  822.021371]  r9:0000000b r8:000def80 r7:00000000 r6:00000000 r5:cb67a3c0 r4:cb67a3c0
> >    [  822.029517] [<c021088c>] (SyS_write) from [<c010d8a0>] (ret_fast_syscall+0x0/0x3c)
> >    [  822.037467]  r9:c5aba000 r8:c010daa8 r7:00000004 r6:b6f0ad58 r5:000def80 r4:0000000b
> >    [  822.045599] Code: e1a0c00d e92dd800 e24cb004 e0801001 (e5910000)
> > 
> >    Without pm_runtime_{get,put}_sync calls in place,
> >    Similar issue come on "mode_show" and "vbus_store" also.
> > 
> >    fix- call pm_runtime_{get,put}_sync before reading/writing
> >    usb mode/vbus from sysfs.
> > 
> >    As sugguested on commit '2d15f69ed5c1c33f283e77ec161578badde33eaf'
> >    Merged all the fix which reslove same bug in the same file.
> >    "
> >      commit '2d15f69ed5c1c33f283e77ec161578badde33eaf'
> >      Author: Saurav Girepunje <saurav.girepunje@gmail.com>
> >      Date:   Thu Aug 8 00:05:03 2019 +0530
> > 
> >      usb: musb: Fix external abort on non-linefetch for vbus_store
> > 
> >      Without pm_runtime_{get,put}_sync calls in place, writing
> >      vbus value via /sys causes the error
> > 
> >      "Unhandled fault external abort on non-linefetch"
> > 
> >        On Thu, Aug 08, 2019 at 12:11:46AM +0530, Saurav Girepunje wrote:
> > 	> Without pm_runtime_{get,put}_sync calls in place, writing
> > 	> vbus value via /sys causes the error
> > 	>
> > 	> "Unhandled fault external abort on non-linefetch"
> > 	>
> > 	> Signed-off-by: Saurav Girepunje <saurav.girepunje@xxxxxxxxx>
> > 
> > 	Thanks for the patch. Can you please merge another two similar patches
> > 	you just sent with this one? They all fix the same bug in the same file
> > 	so could be just in one patch.
> > 
> > 	-Bin.
> >    "
> > 
> > Signed-off-by: Saurav Girepunje <saurav.girepunje@gmail.com>
> > ---
> > drivers/usb/musb/musb_core.c | 6 ++++++
> > 1 file changed, 6 insertions(+)
> > 
> > diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c
> > index bd63450af76a..dc7d786feb58 100644
> > --- a/drivers/usb/musb/musb_core.c
> > +++ b/drivers/usb/musb/musb_core.c
> > @@ -1723,9 +1723,11 @@ mode_show(struct device *dev, struct device_attribute *attr, char *buf)
> > 	unsigned long flags;
> > 	int ret;
> > 
> > +	pm_runtime_get_sync(dev);
> > 	spin_lock_irqsave(&musb->lock, flags);
> > 	ret = sprintf(buf, "%s\n", usb_otg_state_string(musb->xceiv->otg->state));
> > 	spin_unlock_irqrestore(&musb->lock, flags);
> > +	pm_runtime_put_sync(dev);
> > 
> > 	return ret;
> > }
> > @@ -1738,6 +1740,7 @@ mode_store(struct device *dev, struct device_attribute *attr,
> > 	unsigned long	flags;
> > 	int		status;
> > 
> > +	pm_runtime_get_sync(dev);
> > 	spin_lock_irqsave(&musb->lock, flags);
> > 	if (sysfs_streq(buf, "host"))
> > 		status = musb_platform_set_mode(musb, MUSB_HOST);
> > @@ -1748,6 +1751,7 @@ mode_store(struct device *dev, struct device_attribute *attr,
> > 	else
> > 		status = -EINVAL;
> > 	spin_unlock_irqrestore(&musb->lock, flags);
> > +	pm_runtime_put_sync(dev);
> > 
> > 	return (status == 0) ? n : status;
> > }
> > @@ -1766,6 +1770,7 @@ vbus_store(struct device *dev, struct device_attribute *attr,
> > 		return -EINVAL;
> > 	}
> > 
> > +	pm_runtime_get_sync(dev);
> > 	spin_lock_irqsave(&musb->lock, flags);
> > 	/* force T(a_wait_bcon) to be zero/unlimited *OR* valid */
> > 	musb->a_wait_bcon = val ? max_t(int, val, OTG_TIME_A_WAIT_BCON) : 0 ;
> > @@ -1773,6 +1778,7 @@ vbus_store(struct device *dev, struct device_attribute *attr,
> > 		musb->is_active = 0;
> > 	musb_platform_try_idle(musb, jiffies + msecs_to_jiffies(val));
> > 	spin_unlock_irqrestore(&musb->lock, flags);
> > +	pm_runtime_put_sync(dev);
> > 
> > 	return n;
> > }
> > -- 
> > 2.20.1
> > 

  parent reply	other threads:[~2020-02-06 19:17 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20191027083616.GA13761@saurav>
2019-12-11 19:09 ` [PATCH] usb: musb: Fix external abort on non-linefetch Bin Liu
     [not found]   ` <20200119060404.GA104504@google.com>
2020-02-06 19:17     ` Bin Liu [this message]
2020-01-26 16:11 Saurav Girepunje
2020-01-26 16:18 ` Greg KH
  -- strict thread matches above, loose matches on Subject: below --
2019-10-28 19:31 Saurav Girepunje

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200206191735.GA11124@iaqt7 \
    --to=b-liu@ti.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-usb@vger.kernel.org \
    --cc=saurav.girepunje@gmail.com \
    --cc=saurav.girepunje@hotmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).