On Wed, Mar 3, 2021 at 7:50 AM Eric W. Biederman <ebiederm@xxxxxxxxxxxx> wrote:
>
> Ilya Lipnitskiy <ilya.lipnitskiy@xxxxxxxxx> writes:
>
> > On Tue, Mar 2, 2021 at 11:37 AM Eric W. Biederman <ebiederm@xxxxxxxxxxxx> wrote:
> >>
> >> Ilya Lipnitskiy <ilya.lipnitskiy@xxxxxxxxx> writes:
> >>
> >> > On Mon, Mar 1, 2021 at 12:43 PM Eric W. Biederman <ebiederm@xxxxxxxxxxxx> wrote:
> >> >>
> >> >> Ilya Lipnitskiy <ilya.lipnitskiy@xxxxxxxxx> writes:
> >> >>
> >> >> > Eric, All,
> >> >> >
> >> >> > The following error appears when running Linux 5.10.18 on an embedded
> >> >> > MIPS mt7621 target:
> >> >> > [ 0.301219] BUG: Bad rss-counter state mm:(ptrval) type:MM_ANONPAGES val:1
> >> >> >
> >> >> > Being a very generic error, I started digging and added a stack dump
> >> >> > before the BUG:
> >> >> > Call Trace:
> >> >> > [<80008094>] show_stack+0x30/0x100
> >> >> > [<8033b238>] dump_stack+0xac/0xe8
> >> >> > [<800285e8>] __mmdrop+0x98/0x1d0
> >> >> > [<801a6de8>] free_bprm+0x44/0x118
> >> >> > [<801a86a8>] kernel_execve+0x160/0x1d8
> >> >> > [<800420f4>] call_usermodehelper_exec_async+0x114/0x194
> >> >> > [<80003198>] ret_from_kernel_thread+0x14/0x1c
> >> >> >
> >> >> > So that's how I got to looking at fs/exec.c and noticed quite a few
> >> >> > changes last year. Turns out this message only occurs once very early
> >> >> > at boot during the very first call to kernel_execve. current->mm is
> >> >> > NULL at this stage, so acct_arg_size() is effectively a no-op.
> >> >>
> >> >> If you believe this is a new error you could bisect the kernel
> >> >> to see which change introduced the behavior you are seeing.
> >> >>
> >> >> > More digging, and I traced the RSS counter increment to:
> >> >> > [<8015adb4>] add_mm_counter_fast+0xb4/0xc0
> >> >> > [<80160d58>] handle_mm_fault+0x6e4/0xea0
> >> >> > [<80158aa4>] __get_user_pages.part.78+0x190/0x37c
> >> >> > [<8015992c>] __get_user_pages_remote+0x128/0x360
> >> >> > [<801a6d9c>] get_arg_page+0x34/0xa0
> >> >> > [<801a7394>] copy_string_kernel+0x194/0x2a4
> >> >> > [<801a880c>] kernel_execve+0x11c/0x298
> >> >> > [<800420f4>] call_usermodehelper_exec_async+0x114/0x194
> >> >> > [<80003198>] ret_from_kernel_thread+0x14/0x1c
> >> >> >
> >> >> > In fact, I also checked vma_pages(bprm->vma) and lo and behold it is set to 1.
> >> >> >
> >> >> > How is fs/exec.c supposed to handle implied RSS increments that happen
> >> >> > due to page faults when discarding the bprm structure? In this case,
> >> >> > the bug-generating kernel_execve call never succeeded, it returned -2,
> >> >> > but I didn't trace exactly what failed.
> >> >>
> >> >> Unless I am mistaken any left over pages should be purged by exit_mmap
> >> >> which is called by mmput before mmput calls mmdrop.
> >> > Good to know. Some more digging and I can say that we hit this error
> >> > when trying to unmap PFN 0 (is_zero_pfn(pfn) returns TRUE,
> >> > vm_normal_page returns NULL, zap_pte_range does not decrement
> >> > MM_ANONPAGES RSS counter). Is my understanding correct that PFN 0 is
> >> > usable, but special? Or am I totally off the mark here?
> >>
> >> It would be good to know if that is the page that get_user_pages_remote
> >> returned to copy_string_kernel. The zero page that is always zero,
> >> should never be returned when a writable mapping is desired.
> >
> > Indeed, pfn 0 is returned from get_arg_page: (page is 0x809cf000,
> > page_to_pfn(page) is 0) and it is the same page that is being freed and not
> > refcounted in mmput/zap_pte_range. Confirmed with good old printk. Also,
> > ZERO_PAGE(0)==0x809fc000 -> PFN 5120.
> >
> > I think I have found the problem though, after much digging and thanks to all
> > the information provided. init_zero_pfn() gets called too late (after
> > the call to
> > is_zero_pfn(0) from mmput returns true), until then zero_pfn == 0, and after,
> > zero_pfn == 5120. Boom.
> >
> > So PFN 0 is special, but only for a little bit, enough for something
> > on my system
> > to call kernel_execve :)
> >
> > Question: is my system not supposed to be calling kernel_execve this
> > early or does
> > init_zero_pfn() need to happen earlier? init_zero_pfn is currently a
> > core_initcall.
>
> Looking quickly it seems that init_zero_pfn() is in mm/memory.c and is
> common for both mips and x86. Further it appears init_zero_pfn() has
> been that was since 2009 a13ea5b75964 ("mm: reinstate ZERO_PAGE").
>
> Given the testing that x86 gets and that nothing like this has been
> reported it looks like whatever driver is triggering the kernel_execve
> is doing something wrong.
>
> Because honestly. If the zero page isn't working there is not a chance
> that anything in userspace is working so it is clearly much too early.
>
> I suspect there is some driver that is initialized very early that is
> doing something that looks innocuous (like triggering a hotplug event)
> and that happens to cause a call_usermode_helper which then calls
> kernel_execve.
Here is the data that's passed into the very first kernel_execve call:
kernel_filename: /sbin/hotplug
argv: [/sbin/hotplug, bus]
envp: [ACTION=add, DEVPATH=/bus/workqueue, SUBSYSTEM=bus, SEQNUM=4,
HOME=/, PATH=/sbin:/bin:/usr/sbin:/usr/bin]
It comes from kobject_uevent_env() calling call_usermodehelper_exec()
with UMH_NO_WAIT.
Trace:
[<80340dc8>] kobject_uevent_env+0x7e4/0x7ec
[<8033f8b8>] kset_register+0x68/0x88
[<803cf824>] bus_register+0xdc/0x34c
[<803cfac8>] subsys_virtual_register+0x34/0x78
[<8086afb0>] wq_sysfs_init+0x1c/0x4c
[<80001648>] do_one_initcall+0x50/0x1a8
[<8086503c>] kernel_init_freeable+0x230/0x2c8
[<8066bca0>] kernel_init+0x10/0x100
[<80003038>] ret_from_kernel_thread+0x14/0x1c
A bunch of other bus devices are initialized at the same time, but
SEQNUM=4 gets to go first for some reason:
[ 0.420497] smp: Brought up 1 node, 4 CPUs
[ 0.431204] ACTION:add DEVPATH:/bus/platform SUBSYSTEM:bus SEQNUM: 1
[ 0.431249] ACTION:add DEVPATH:/bus/cpu SUBSYSTEM:bus SEQNUM: 2
[ 0.440594] ACTION:add DEVPATH:/bus/container SUBSYSTEM:bus SEQNUM: 3
[ 0.449994] ACTION:add DEVPATH:/bus/workqueue SUBSYSTEM:bus SEQNUM: 4
Since both wq_sysfs_init() and init_zero_pfn() are annotated with
core_initcall() is there a race?
Maybe there is still an argument for moving init_zero_pfn() to
early_initcall()? According to the comment above init_zero_pfn(),
"CONFIG_MMU architectures set up ZERO_PAGE in their paging_init()".
paging_init() gets called in setup_arch(), which is way before
do_pre_smp_initcalls(), so it should work, right? Obviously something
that needs to be tested, but are my assumptions correct?
FWIW I tested it on my MIPS device and it boots fine and the BUG
message is gone. I still don't know why it started appearing on 5.10+,
maybe some core_initcalls got added that made the race worse?
Ilya
