Commit graph

1,044,825 commits

Author SHA1 Message Date
Arnd Bergmann
3d9d00bd18 net: bonding: move ioctl handling to private ndo operation
All other user triggered operations are gone from ndo_ioctl, so move
the SIOCBOND family into a custom operation as well.

The .ndo_ioctl() helper is no longer called by the dev_ioctl.c code now,
but there are still a few definitions in obsolete wireless drivers as well
as the appletalk and ieee802154 layers to call SIOCSIFADDR/SIOCGIFADDR
helpers from inside the kernel.

Cc: Jay Vosburgh <j.vosburgh@gmail.com>
Cc: Veaceslav Falico <vfalico@gmail.com>
Cc: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:45 +01:00
Arnd Bergmann
ad2f99aedf net: bridge: move bridge ioctls out of .ndo_do_ioctl
Working towards obsoleting the .ndo_do_ioctl operation entirely,
stop passing the SIOCBRADDIF/SIOCBRDELIF device ioctl commands
into this callback.

My first attempt was to add another ndo_siocbr() callback, but
as there is only a single driver that takes these commands and
there is already a hook mechanism to call directly into this
driver, extend this hook instead, and use it for both the
deviceless and the device specific ioctl commands.

Cc: Roopa Prabhu <roopa@nvidia.com>
Cc: Nikolay Aleksandrov <nikolay@nvidia.com>
Cc: bridge@lists.linux-foundation.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:45 +01:00
Arnd Bergmann
88fc023f7d net: socket: return changed ifreq from SIOCDEVPRIVATE
Some drivers that use SIOCDEVPRIVATE ioctl commands modify
the ifreq structure and expect it to be passed back to user
space, which has never really happened for compat mode
because the calling these drivers through ndo_do_ioctl
requires overwriting the ifr_data pointer.

Now that all drivers are converted to ndo_siocdevprivate,
change it to handle this correctly in both compat and
native mode.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:45 +01:00
Arnd Bergmann
ad7eab2ab0 net: split out ndo_siowandev ioctl
In order to further reduce the scope of ndo_do_ioctl(), move
out the SIOCWANDEV handling into a new network device operation
function.

Adjust the prototype to only pass the if_settings sub-structure
in place of the ifreq, and remove the redundant 'cmd' argument
in the process.

Cc: Krzysztof Halasa <khc@pm.waw.pl>
Cc: "Jan \"Yenya\" Kasprzak" <kas@fi.muni.cz>
Cc: Kevin Curtis <kevin.curtis@farsite.co.uk>
Cc: Zhao Qiang <qiang.zhao@nxp.com>
Cc: Martin Schiller <ms@dev.tdt.de>
Cc: Jiri Slaby <jirislaby@kernel.org>
Cc: linux-x25@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:45 +01:00
Arnd Bergmann
a76053707d dev_ioctl: split out ndo_eth_ioctl
Most users of ndo_do_ioctl are ethernet drivers that implement
the MII commands SIOCGMIIPHY/SIOCGMIIREG/SIOCSMIIREG, or hardware
timestamping with SIOCSHWTSTAMP/SIOCGHWTSTAMP.

Separate these from the few drivers that use ndo_do_ioctl to
implement SIOCBOND, SIOCBR and SIOCWANDEV commands.

This is a purely cosmetic change intended to help readers find
their way through the implementation.

Cc: Doug Ledford <dledford@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Jay Vosburgh <j.vosburgh@gmail.com>
Cc: Veaceslav Falico <vfalico@gmail.com>
Cc: Andy Gospodarek <andy@greyhouse.net>
Cc: Andrew Lunn <andrew@lunn.ch>
Cc: Vivien Didelot <vivien.didelot@gmail.com>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Cc: Vladimir Oltean <olteanv@gmail.com>
Cc: Leon Romanovsky <leon@kernel.org>
Cc: linux-rdma@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:45 +01:00
Arnd Bergmann
a554bf96b4 dev_ioctl: pass SIOCDEVPRIVATE data separately
The compat handlers for SIOCDEVPRIVATE are incorrect for any driver that
passes data as part of struct ifreq rather than as an ifr_data pointer, or
that passes data back this way, since the compat_ifr_data_ioctl() helper
overwrites the ifr_data pointer and does not copy anything back out.

Since all drivers using devprivate commands are now converted to the
new .ndo_siocdevprivate callback, fix this by adding the missing piece
and passing the pointer separately the whole way.

This further unifies the native and compat logic for socket ioctls,
as the new code now passes the correct pointer as well as the correct
data for both native and compat ioctls.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
8fb75b79cd wan: cosa: remove dead cosa_net_ioctl() function
The ndo_do_ioctl callback is never called with the COSAIO* commands,
so this is never used. Call the hdlc_ioctl function directly instead.

Any user space code that relied on this function working as intended
has never worked in a mainline kernel since before linux-1.0.

Cc: "Jan \"Yenya\" Kasprzak" <kas@fi.muni.cz>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
73d74f61a5 wan: use ndo_siocdevprivate
The wan drivers each support some custom SIOCDEVPRIVATE
ioctls, plus the common SIOCWANDEV command.

Split these so the ioctl callback only deals with SIOCWANDEV
and the rest is handled by ndo_siocdevprivate.

It might make sense to also split out SIOCWANDEV into a
separate callback in order to eventually remove ndo_do_ioctl
entirely.

Cc: Krzysztof Halasa <khc@pm.waw.pl>
Cc: Kevin Curtis <kevin.curtis@farsite.co.uk>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
34f7cac07c ppp: use ndo_siocdevprivate
ppp has a custom statistics interface using SIOCDEVPRIVATE
ioctl commands that works correctly in compat mode.

Convert it to use ndo_siocdevprivate as a cleanup.

Cc: Paul Mackerras <paulus@samba.org>
Cc: linux-ppp@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
cc0aa831a0 sb1000: use ndo_siocdevprivate
The private sb1000 ioctl commands all work correctly in
compat mode. Change the to ndo_siocdevprivate as a cleanup.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
81a68110a2 hippi: use ndo_siocdevprivate
The rr_ioctl uses private ioctl commands that correctly pass
all data through ifr_data, which works fine in compat mode.

Change it to use ndo_siocdevprivate as a cleanup.

Cc: Jes Sorensen <jes@trained-monkey.org>
Cc: linux-hippi@sunsite.dk
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
3e7a1c7c56 ip_tunnel: use ndo_siocdevprivate
The various ipv4 and ipv6 tunnel drivers each implement a set
of 12 SIOCDEVPRIVATE commands for managing tunnels. These
all work correctly in compat mode.

Move them over to the new .ndo_siocdevprivate operation.

Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: David Ahern <dsahern@kernel.org>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
ae6af0120d airo: use ndo_siocdevprivate
The airo driver overloads SIOCDEVPRIVATE ioctls with another
set based on SIOCIWFIRSTPRIV. Only the first ones actually
work (also in compat mode) as the others do not get passed
down any more.

Change it over to ndo_siocdevprivate for clarification.

Cc: Kalle Valo <kvalo@codeaurora.org>
Cc: linux-wireless@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
25ec92fbdd hamradio: use ndo_siocdevprivate
hamradio uses a set of private ioctls that do seem to work
correctly in compat mode, as they only rely on the ifr_data
pointer.

Move them over to the ndo_siocdevprivate callback as a cleanup.

Cc: Thomas Sailer <t.sailer@alumni.ethz.ch>
Cc: Joerg Reuter <jreuter@yaina.de>
Cc: Jean-Paul Roubelat <jpr@f6fbb.org>
Cc: linux-hams@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
ebb4a911e0 cxgb3: use ndo_siocdevprivate
cxgb3 has a private multiplexor that works correctly in compat
mode, split out the siocdevprivate callback from do_ioctl for
simplification.

Cc: Raju Rangoju <rajur@chelsio.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
18787eeebd qeth: use ndo_siocdevprivate
qeth has both standard MII ioctls and custom SIOCDEVPRIVATE ones,
all of which work correctly with compat user space.

Move the private ones over to the new ndo_siocdevprivate callback.

Cc: Julian Wiedmann <jwi@linux.ibm.com>
Cc: Karsten Graul <kgraul@linux.ibm.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: linux-s390@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
76b5878cff slip/plip: use ndo_siocdevprivate
slip and plip both use a couple of SIOCDEVPRIVATE ioctl
commands that overload the ifreq layout in a way that is
incompatible with compat mode.

Convert to use ndo_siocdevprivate to allow passing the
data this way, but return an error in compat mode anyway
because the private structure is still incompatible.

This could be fixed as well to make compat work properly.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:44 +01:00
Arnd Bergmann
ef1b5b0c30 net: usb: use ndo_siocdevprivate
The pegasus and rtl8150 drivers use SIOCDEVPRIVATE ioctls
to access their MII registers, in place of the normal
commands. This is broken for all compat ioctls today.

Change to ndo_siocdevprivate to fix it.

Cc: Petko Manolov <petkan@nucleusys.com>
Cc: linux-usb@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
043393d8b4 fddi: use ndo_siocdevprivate
The skfddi driver has a private ioctl and passes the data correctly
through ifr_data, but the use of a pointer in s_skfp_ioctl is
broken in compat mode.

Change the driver to use ndo_siocdevprivate and disallow calling
it in compat mode until a conversion handler is added.

Cc: "Maciej W. Rozycki" <macro@orcam.me.uk>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
d92f7b59d3 eql: use ndo_siocdevprivate
The private ioctls in eql pass the arguments correctly through ifr_data,
but the slaving_request_t and slave_config_t structures are incompatible
with compat mode and need special conversion code in the driver.

Convert to siocdevprivate for now, and return an error when called
in compat mode.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
32d05468c4 tehuti: use ndo_siocdevprivate
Tehuti only implements private ioctl commands, and implements
them by overriding the ifreq layout, which is broken in
compat mode.

Move it to the ndo_siocdevprivate callback in order to fix this.

Cc: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
99b78a37a3 hamachi: use ndo_siocdevprivate
hamachi has one command that overloads the ifreq argument
and requires a conversion to ndo_siocdevprivate in order to
make compat mode work, so split it from ndo_ioctl.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
dbecb011eb appletalk: use ndo_siocdevprivate
appletalk has three SIOCDEVPRIVATE ioctl commands that are
broken in compat mode because the passed structure contains
a pointer.

Change it over to ndo_siocdevprivate for consistency and
make it return an error when called in compat mode. This
could be fixed if there are still users.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
232ec98ec3 bonding: use siocdevprivate
The bonding driver supports two command codes for each operation: one
in the SIOCDEVPRIVATE range and another one with the same definition
but a unique command code.

Only the second set currently works in compat mode, as the ifr_data
expansion overwrites part of the ifr_slave field.

Move the private ones into ndo_siocdevprivate and change the
implementation to call the other function.  This makes both version
work correctly.

Cc: Jay Vosburgh <j.vosburgh@gmail.com>
Cc: Veaceslav Falico <vfalico@gmail.com>
Cc: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
029a4fef6b tulip: use ndo_siocdevprivate
The tulip driver has a debugging method over ioctl built-in, but it
does not actually check the command type, which may end up leading
to random behavior when trying to run other ioctls on it.

Change the driver to use ndo_siocdevprivate and limit the execution
further to the first private command code. If anyone still has tools
to run these debugging commands, they might have to be patched for
it if they pass different ioctl command.

The function has existed in this form since the driver was merged in
Linux-1.1.86.

Cc: linux-parisc@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
4747c1a8bc phonet: use siocdevprivate
phonet has a single private ioctl that is broken in compat
mode on big-endian machines today because the data returned
from it is never copied back to user space.

Move it over to the ndo_siocdevprivate callback, which also
fixes the compat issue.

Cc: Remi Denis-Courmont <courmisch@gmail.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Rémi Denis-Courmont <courmisch@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
561d835281 bridge: use ndo_siocdevprivate
The bridge driver has an old set of ioctls using the SIOCDEVPRIVATE
namespace that have never worked in compat mode and are explicitly
forbidden already.

Move them over to ndo_siocdevprivate and fix compat mode for these,
because we can.

Cc: Roopa Prabhu <roopa@nvidia.com>
Cc: Nikolay Aleksandrov <nikolay@nvidia.com>
Cc: bridge@lists.linux-foundation.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
3f3fa53407 hostap: use ndo_siocdevprivate
hostap has a combination of iwpriv ioctls that do not work at
all, and two SIOCDEVPRIVATE commands that work natively but
lack a compat conversion handler.

For the moment, move them over to the new ndo_siocdevprivate
interface and return an error for compat mode.

Cc: Jouni Malinen <j@w1.fi>
Cc: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
3343c49a95 staging: wlan-ng: use siocdevprivate
wlan-ng has two private ioctls that correctly work in compat
mode. Move these over to the new ndo_siocdevprivate mechanism.

The p80211netdev_ethtool() function is commented out and
has no use here, so this can be removed

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
89939e8906 staging: rtlwifi: use siocdevprivate
rtl8188eu has an "android private" ioctl command multiplexer
that is not currently safe for use in compat mode because
of its triple-indirect pointer.

rtl8723bs uses a different interface on the SIOCDEVPRIVATE
command, based on the iwpriv data structure

Both also have normal unreachable iwpriv commands, and all
of the above should probably just get removed. For the
moment, just switch over to the new interface.

Cc: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
Arnd Bergmann
b9067f5dc4 net: split out SIOCDEVPRIVATE handling from dev_ioctl
SIOCDEVPRIVATE ioctl commands are mainly used in really old
drivers, and they have a number of problems:

- They hide behind the normal .ndo_do_ioctl function that
  is also used for other things in modern drivers, so it's
  hard to spot a driver that actually uses one of these

- Since drivers use a number different calling conventions,
  it is impossible to support compat mode for them in
  a generic way.

- With all drivers using the same 16 commands codes, there
  is no way to introspect the data being passed through
  things like strace.

Add a new net_device_ops callback pointer, to address the
first two of these. Separating them from .ndo_do_ioctl
makes it easy to grep for drivers with a .ndo_siocdevprivate
callback, and the unwieldy name hopefully makes it easier
to spot in code review.

By passing the ifreq structure and the ifr_data pointer
separately, it is no longer necessary to overload these,
and the driver can use either one for a given command.

Cc: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:11:43 +01:00
David S. Miller
2fba2eae30 Merge branch 'tcp-rack'
Neal Cardwell says:

====================
more accurate DSACK processing for RACK-TLP

This patch series includes two minor improvements to tighten up the accuracy of
the processing of incoming DSACK information, so that RACK-TLP behavior is
faster and more precise: first, to ensure we detect packet loss in some extra
corner cases; and second, to avoid growing the RACK reordering window (and
delaying fast recovery) in cases where it seems clear we don't need to.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:09:29 +01:00
Neal Cardwell
a657db0350 tcp: more accurately check DSACKs to grow RACK reordering window
Previously, a DSACK could expand the RACK reordering window when no
reordering has been seen, and/or when the DSACK was due to an
unnecessary TLP retransmit (rather than a spurious fast recovery due
to reordering). This could result in unnecessarily growing the RACK
reordering window and thus unnecessarily delaying RACK-based fast
recovery episodes.

To avoid these issues, this commit tightens the conditions under which
a DSACK triggers the RACK reordering window to grow, so that a
connection only expands its RACK reordering window if:

(a) reordering has been seen in the connection
(b) a DSACKed range does not match the most recent TLP retransmit

Signed-off-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:07:21 +01:00
Yuchung Cheng
63f367d9de tcp: more accurately detect spurious TLP probes
Previously TLP is considered spurious if the sender receives any
DSACK during a TLP episode. This patch further checks the DSACK
sequences match the TLP's to improve accuracy.

Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-27 20:07:21 +01:00
Thomas Gleixner
bb7262b295 timers: Move clearing of base::timer_running under base:: Lock
syzbot reported KCSAN data races vs. timer_base::timer_running being set to
NULL without holding base::lock in expire_timers().

This looks innocent and most reads are clearly not problematic, but
Frederic identified an issue which is:

 int data = 0;

 void timer_func(struct timer_list *t)
 {
    data = 1;
 }

 CPU 0                                            CPU 1
 ------------------------------                   --------------------------
 base = lock_timer_base(timer, &flags);           raw_spin_unlock(&base->lock);
 if (base->running_timer != timer)                call_timer_fn(timer, fn, baseclk);
   ret = detach_if_pending(timer, base, true);    base->running_timer = NULL;
 raw_spin_unlock_irqrestore(&base->lock, flags);  raw_spin_lock(&base->lock);

 x = data;

If the timer has previously executed on CPU 1 and then CPU 0 can observe
base->running_timer == NULL and returns, assuming the timer has completed,
but it's not guaranteed on all architectures. The comment for
del_timer_sync() makes that guarantee. Moving the assignment under
base->lock prevents this.

For non-RT kernel it's performance wise completely irrelevant whether the
store happens before or after taking the lock. For an RT kernel moving the
store under the lock requires an extra unlock/lock pair in the case that
there is a waiter for the timer, but that's not the end of the world.

Reported-by: syzbot+aa7c2385d46c5eba0b89@syzkaller.appspotmail.com
Reported-by: syzbot+abea4558531bae1ba9fe@syzkaller.appspotmail.com
Fixes: 030dcdd197 ("timers: Prepare support for PREEMPT_RT")
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://lore.kernel.org/r/87lfea7gw8.fsf@nanos.tec.linutronix.de
Cc: stable@vger.kernel.org
2021-07-27 20:57:44 +02:00
Liam Beguin
284c537a8a clk: lmk04832: drop redundant fallthrough statements
When the body of a case statement is empty, it is well understood that
it is intentional and explicit fallthrough statements are not required.
Drop them.

Signed-off-by: Liam Beguin <liambeguin@gmail.com>
Link: https://lore.kernel.org/r/20210708211645.3621902-1-liambeguin@gmail.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
2021-07-27 11:52:30 -07:00
Paul E. McKenney
87859a8e3f tools/memory-model: Document data_race(READ_ONCE())
It is possible to cause KCSAN to ignore marked accesses by applying
__no_kcsan to the function or applying data_race() to the marked accesses.
These approaches allow the developer to restrict compiler optimizations
while also causing KCSAN to ignore diagnostic accesses.

This commit therefore updates the documentation accordingly.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:48:55 -07:00
Manfred Spraul
f92975d76d tools/memory-model: Heuristics using data_race() must handle all values
Data loaded for use by some sorts of heuristics can tolerate the
occasional erroneous value.  In this case the loads may use data_race()
to give the compiler full freedom to optimize while also informing KCSAN
of the intent.  However, for this to work, the heuristic needs to be
able to tolerate any erroneous value that could possibly arise.  This
commit therefore adds a paragraph spelling this out.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:48:55 -07:00
Paul E. McKenney
436eef23c4 tools/memory-model: Add example for heuristic lockless reads
This commit adds example code for heuristic lockless reads, based loosely
on the sem_lock() and sem_unlock() functions.

[ paulmck: Apply Alan Stern and Manfred Spraul feedback. ]

Reported-by: Manfred Spraul <manfred@colorfullife.com>
[ paulmck: Update per Manfred Spraul and Hillf Danton feedback. ]
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:47:34 -07:00
Miles Chen
d17e4e62df clk: mediatek: make COMMON_CLK_MT8167* depend on COMMON_CLK_MT8167
I found that COMMON_CLK_MT8167* do not depend on COMMON_CLK_MT8167,
so it is possible to config:

CONFIG_COMMON_CLK_MT8167=n
CONFIG_COMMON_CLK_MT8167_*=y

Although it does not cause build breaks with such configuration,
I think it is clearer to make COMMON_CLK_MT8167* depend on
COMMON_CLK_MT8167.

Signed-off-by: Miles Chen <miles.chen@mediatek.com>
Link: https://lore.kernel.org/r/20210716051732.3422-1-miles.chen@mediatek.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
2021-07-27 11:46:27 -07:00
Paul E. McKenney
06ca91448f torture: Make kvm-test-1-run-qemu.sh check for reboot loops
It turns out that certain types of early boot bugs can result in reboot
loops, even within a guest OS running under qemu/KVM.  This commit
therefore upgrades the kvm-test-1-run-qemu.sh script's hang-detection
heuristics to detect such situations and to terminate the run when
they occur.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:33 -07:00
Paul E. McKenney
5f33809ec2 torture: Add timestamps to kvm-test-1-run-qemu.sh output
The kvm-test-1-run-qemu.sh script logs the torture-test start time and
also when it starts getting impatient for the test to finish.  However, it
does not timestamp these log messages, which can make debugging needlessly
challenging.  This commit therefore adds timestamps to these messages.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:33 -07:00
Paul E. McKenney
b3bf9632ef torture: Don't use "test" command's "-a" argument
There was a time long ago when the "test" command's documentation
claimed that the "-a" and "-o" arguments did something useful.
But this documentation now suggests letting the shell execute
these boolean operators, so this commit applies that suggestion to
kvm-test-1-run-qemu.sh.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:33 -07:00
Paul E. McKenney
a5202e173d torture: Make kvm-test-1-run-batch.sh select per-scenario affinity masks
This commit causes kvm-test-1-run-batch.sh to use the new
kvm-assign-cpus.sh and kvm-get-cpus-script.sh scripts to create a
TORTURE_AFFINITY environment variable containing either an empty string
(for no affinity) or a list of CPUs to pin the scenario's vCPUs to.
The additional change to kvm-test-1-run.sh places the per-scenario
number-of-CPUs information where it can easily be found.

If there is some reason why affinity cannot be supplied, this commit
prints and logs the reason via changes to kvm-again.sh.

Finally, this commit updates the kvm-remote.sh script to copy the
qemu-affinity output files back to the host system.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:33 -07:00
Paul E. McKenney
9e528a84c9 torture: Consistently name "qemu*" test output files
There is "qemu-affinity", "qemu-cmd", "qemu-retval", but also "qemu_pid".
This is hard to remember, not so good for bash tab completion, and just
plain inconsistent.  This commit therefore renames the "qemu_pid" file to
"qemu-pid".  A couple of the scripts must deal with old runs, and thus
must handle both "qemu_pid" and "qemu-pid", but new runs will produce
"qemu-pid".

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:33 -07:00
Paul E. McKenney
4567c76a8e torture: Use numeric taskset argument in jitter.sh
The jitter.sh script has some entertaining awk code to generate a
hex mask from a randomly selected CPU number, which is handed to the
"taskset" command.  Except that this command has a "-c" parameter to
take a comma/dash-separated list of CPU numbers.  This commit therefore
saves a few lines of awk by switching to a single-number CPU list.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:32 -07:00
Paul E. McKenney
de2909461c rcutorture: Upgrade two-CPU scenarios to four CPUs
There is no way to place the vCPUs in a two-CPU rcutorture scenario to
get variable memory latency.  This commit therefore upgrades the current
two-CPU rcutorture scenarios to four CPUs.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:32 -07:00
Paul E. McKenney
bdf5ca1201 torture: Make kvm-test-1-run-qemu.sh apply affinity
This commit causes the kvm-test-1-run-qemu.sh script to check the
TORTURE_AFFINITY environment variable and to add "taskset" commands to
the qemu-cmd file.  The first "taskset" command is applied only if the
TORTURE_AFFINITY environment variable is a non-empty string, and this
command pins the current scenario's guest OS to the specified CPUs.
The second "taskset" command reports the guest OS's affinity in a new
"qemu-affinity" file.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:32 -07:00
Paul E. McKenney
8220a11849 torture: Don't redirect qemu-cmd comment lines
Currently, kvm-test-1-run-qemu.sh applies redirection to each and every
line of each qemu-cmd script.  Only the first line (the only one that
is not a bash comment) needs to be redirected.  Although redirecting
the comments is currently harmless, just adding to the comment, it is
an accident waiting to happen.  This commit therefore adjusts the "sed"
command to redirect only the qemu-system* command itself.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:41:32 -07:00
Paul E. McKenney
cdeef67d8f torture: Make kvm.sh select per-scenario affinity masks
This commit causes kvm.sh to use the new kvm-assign-cpus.sh and
kvm-get-cpus-script.sh scripts to create a TORTURE_AFFINITY environment
variable containing either an empty string (for no affinity) or a list
of CPUs to pin the scenario's vCPUs to.  A later commit will make
use of this information to actually pin the vCPUs.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
2021-07-27 11:40:30 -07:00