edac: edac_device_sysfs: reject poll_msec value 0
PATCH 1/2 edac: edac_device_sysfs: reject poll_msec value 0
The comment says it must be at least 1ms but there was never a check. Writing 0 sets the workqueue delay to 0 jiffies, which makes it spin.
--- drivers/edac/edac_device_sysfs.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/edac/edac_device_sysfs.c +4 -0
@@ -117,6 +117,10 @@ static ssize_t edac_device_ctl_poll_msec_store(struct edac_device_ctl_info ret =kstrtoul (data,0 , &value);if (ret <0 )return ret; + +if (value <1 ) +return -EINVAL ; +edac_device_reset_delay_period (ctl_info, value);return count;
Signature · 1 line
2.55.0
PATCH 2/2 edac: edac_device: don't re-arm workqueue during teardown
edac_device_del_device stops the workqueue then removes the sysfs attributes. But the sysfs store for poll_msec calls edac_device_reset_delay_period which unconditionally re-arms the workqueue. If someone writes to poll_msec in that window, the re-armed work runs after edac_dev is freed.
Check op_state instead. del_device already sets it to OP_OFFLINE before tearing down the workqueue, so reset_delay_period will see it and skip the mod_work call.
--- drivers/edac/edac_device.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/edac/edac_device.c +4 -0
@@ -354,6 +354,10 @@ void edac_device_reset_delay_period(struct edac_device_ctl_info *edac_dev, edac_dev->poll_msec = msec; edac_dev->delay =msecs_to_jiffies (msec); +/* Don't re-arm the workqueue if the device is being torn down */ +if (edac_dev->op_state ==OP_OFFLINE ) +return ; +/* See comment in edac_device_workq_setup() above */ if (edac_dev->poll_msec ==DEFAULT_POLL_INTERVAL )edac_mod_work (&edac_dev->work,round_jiffies_relative (edac_dev->delay));
Signature · 1 line
2.55.0
Borislav Petkov
maintainerQuoted 1 line · Jad Keskes
Subject: Re: [PATCH 1/2] edac:
edac_device_sysfs: rejectpoll_msecvalue 0
Do
$ git log -p drivers/edac/
to get an idea about the commit title and formatting the EDAC tree does. In this case:
EDAC/device_sysfs: Reject poll_msec value 0
Quoted 2 lines
The comment says it must be at least 1ms but there was never a check. Writing 0 sets the workqueue delay to 0 jiffies, which makes it spin.
You might run this through AI to get better/more informative formulations.
Some hints about a good commit message:
https://docs.kernel.org/process/submitting-patches.html
drivers/edac/edac_device_sysfs.c static ssize_t edac_device_ctl_poll_msec_store(struct edac_device_ctl_info · line 117
Quoted 11 lines
Signed-off-by Jad Keskes <[email protected]>--- drivers/edac/edac_device_sysfs.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/edac/edac_device_sysfs.c +0
-0@@ -117,6 +117,10 @@ static ssize_t edac_device_ctl_poll_msec_store(struct edac_device_ctl_info ret =kstrtoul (data,0 , &value);
If you do those patches ontop of your previous ones, it doesn't apply:
checking file drivers/edac/edac_device_sysfs.c
Hunk #1 FAILED at 117.
1 out of 1 hunk FAILED
But you need to do those fixes first and then the conversion to kstrtoul() because the fixes will go to stable while the kstrtoul() conversion is not stable material.
Which means, you'd need to add a Fixes: tag to those but that's fine if you can't find which patch causes this - I believe this has always been this way.
So, how about a new patchset with all that review feedback addressed, the fixes first, the conversion following, all nicely tested, if you don't know how use this here in a VM:
https://lore.kernel.org/all/[email protected]
Thanks.
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquette