When user reads file 'trace_pipe', kernel keeps printing following logs
that warn at "cpu_buffer->reader_page->read > rb_page_size(reader)" in
rb_get_reader_page(). It just looks like there's an infinite loop in
tracing_read_pipe(). This problem occurs several times on arm64 platform
when testing v5.10 and below.
Call trace:
rb_get_reader_page+0x248/0x1300
rb_buffer_peek+0x34/0x160
ring_buffer_peek+0xbc/0x224
peek_next_entry+0x98/0xbc
__find_next_entry+0xc4/0x1c0
trace_find_next_entry_inc+0x30/0x94
tracing_read_pipe+0x198/0x304
vfs_read+0xb4/0x1e0
ksys_read+0x74/0x100
__arm64_sys_read+0x24/0x30
el0_svc_common.constprop.0+0x7c/0x1bc
do_el0_svc+0x2c/0x94
el0_svc+0x20/0x30
el0_sync_handler+0xb0/0xb4
el0_sync+0x160/0x180
Then I dump the vmcore and look into the problematic per_cpu ring_buffer,
I found that tail_page/commit_page/reader_page are on the same page while
reader_page->read is obviously abnormal:
tail_page == commit_page == reader_page == {
.write = 0x100d20,
.read = 0x8f9f4805, // Far greater than 0xd20, obviously abnormal!!!
.entries = 0x10004c,
.real_end = 0x0,
.page = {
.time_stamp = 0x857257416af0,
.commit = 0xd20, // This page hasn't been full filled.
// .data[0...0xd20] seems normal.
}
}
The root cause is most likely the race that reader and writer are on the
same page while reader saw an event that not fully committed by writer.
To fix this, add memory barriers to make sure the reader can see the
content of what is committed. Since commit a0fcaaed0c ("ring-buffer: Fix
race between reset page and reading page") has added the read barrier in
rb_get_reader_page(), here we just need to add the write barrier.
Link: https://lore.kernel.org/linux-trace-kernel/20230325021247.2923907-1-zhengyejian1@huawei.com
Cc: stable@vger.kernel.org
Fixes: 77ae365eca ("ring-buffer: make lockless")
Suggested-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Currently, the "last_cmd" variable can be accessed by multiple processes
asynchronously when multiple users manipulate synthetic_events node
at the same time, it could lead to use-after-free or double-free.
This patch add "lastcmd_mutex" to prevent "last_cmd" from being accessed
asynchronously.
================================================================
It's easy to reproduce in the KASAN environment by running the two
scripts below in different shells.
script 1:
while :
do
echo -n -e '\x88' > /sys/kernel/tracing/synthetic_events
done
script 2:
while :
do
echo -n -e '\xb0' > /sys/kernel/tracing/synthetic_events
done
================================================================
double-free scenario:
process A process B
------------------- ---------------
1.kstrdup last_cmd
2.free last_cmd
3.free last_cmd(double-free)
================================================================
use-after-free scenario:
process A process B
------------------- ---------------
1.kstrdup last_cmd
2.free last_cmd
3.tracing_log_err(use-after-free)
================================================================
Appendix 1. KASAN report double-free:
BUG: KASAN: double-free in kfree+0xdc/0x1d4
Free of addr ***** by task sh/4879
Call trace:
...
kfree+0xdc/0x1d4
create_or_delete_synth_event+0x60/0x1e8
trace_parse_run_command+0x2bc/0x4b8
synth_events_write+0x20/0x30
vfs_write+0x200/0x830
...
Allocated by task 4879:
...
kstrdup+0x5c/0x98
create_or_delete_synth_event+0x6c/0x1e8
trace_parse_run_command+0x2bc/0x4b8
synth_events_write+0x20/0x30
vfs_write+0x200/0x830
...
Freed by task 5464:
...
kfree+0xdc/0x1d4
create_or_delete_synth_event+0x60/0x1e8
trace_parse_run_command+0x2bc/0x4b8
synth_events_write+0x20/0x30
vfs_write+0x200/0x830
...
================================================================
Appendix 2. KASAN report use-after-free:
BUG: KASAN: use-after-free in strlen+0x5c/0x7c
Read of size 1 at addr ***** by task sh/5483
sh: CPU: 7 PID: 5483 Comm: sh
...
__asan_report_load1_noabort+0x34/0x44
strlen+0x5c/0x7c
tracing_log_err+0x60/0x444
create_or_delete_synth_event+0xc4/0x204
trace_parse_run_command+0x2bc/0x4b8
synth_events_write+0x20/0x30
vfs_write+0x200/0x830
...
Allocated by task 5483:
...
kstrdup+0x5c/0x98
create_or_delete_synth_event+0x80/0x204
trace_parse_run_command+0x2bc/0x4b8
synth_events_write+0x20/0x30
vfs_write+0x200/0x830
...
Freed by task 5480:
...
kfree+0xdc/0x1d4
create_or_delete_synth_event+0x74/0x204
trace_parse_run_command+0x2bc/0x4b8
synth_events_write+0x20/0x30
vfs_write+0x200/0x830
...
Link: https://lore.kernel.org/linux-trace-kernel/20230321110444.1587-1-Tze-nan.Wu@mediatek.com
Fixes: 27c888da98 ("tracing: Remove size restriction on synthetic event cmd error logging")
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: "Tom Zanussi" <zanussi@kernel.org>
Signed-off-by: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
When BPF_TRAMP_F_CALL_ORIG is set, BPF trampoline uses BLR to jump
back to the instruction next to call site to call the patched function.
For BTI-enabled kernel, the instruction next to call site is usually
PACIASP, in this case, it's safe to jump back with BLR. But when
the call site is not followed by a PACIASP or bti, a BTI exception
is triggered.
Here is a fault log:
Unhandled 64-bit el1h sync exception on CPU0, ESR 0x0000000034000002 -- BTI
CPU: 0 PID: 263 Comm: test_progs Tainted: GF
Hardware name: linux,dummy-virt (DT)
pstate: 40400805 (nZcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=-c)
pc : bpf_fentry_test1+0xc/0x30
lr : bpf_trampoline_6442573892_0+0x48/0x1000
sp : ffff80000c0c3a50
x29: ffff80000c0c3a90 x28: ffff0000c2e6c080 x27: 0000000000000000
x26: 0000000000000000 x25: 0000000000000000 x24: 0000000000000050
x23: 0000000000000000 x22: 0000ffffcfd2a7f0 x21: 000000000000000a
x20: 0000ffffcfd2a7f0 x19: 0000000000000000 x18: 0000000000000000
x17: 0000000000000000 x16: 0000000000000000 x15: 0000ffffcfd2a7f0
x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000
x11: 0000000000000000 x10: ffff80000914f5e4 x9 : ffff8000082a1528
x8 : 0000000000000000 x7 : 0000000000000000 x6 : 0101010101010101
x5 : 0000000000000000 x4 : 00000000fffffff2 x3 : 0000000000000001
x2 : ffff8001f4b82000 x1 : 0000000000000000 x0 : 0000000000000001
Kernel panic - not syncing: Unhandled exception
CPU: 0 PID: 263 Comm: test_progs Tainted: GF
Hardware name: linux,dummy-virt (DT)
Call trace:
dump_backtrace+0xec/0x144
show_stack+0x24/0x7c
dump_stack_lvl+0x8c/0xb8
dump_stack+0x18/0x34
panic+0x1cc/0x3ec
__el0_error_handler_common+0x0/0x130
el1h_64_sync_handler+0x60/0xd0
el1h_64_sync+0x78/0x7c
bpf_fentry_test1+0xc/0x30
bpf_fentry_test1+0xc/0x30
bpf_prog_test_run_tracing+0xdc/0x2a0
__sys_bpf+0x438/0x22a0
__arm64_sys_bpf+0x30/0x54
invoke_syscall+0x78/0x110
el0_svc_common.constprop.0+0x6c/0x1d0
do_el0_svc+0x38/0xe0
el0_svc+0x30/0xd0
el0t_64_sync_handler+0x1ac/0x1b0
el0t_64_sync+0x1a0/0x1a4
Kernel Offset: disabled
CPU features: 0x0000,00034c24,f994fdab
Memory Limit: none
And the instruction next to call site of bpf_fentry_test1 is ADD,
not PACIASP:
<bpf_fentry_test1>:
bti c
nop
nop
add w0, w0, #0x1
paciasp
For BPF prog, JIT always puts a PACIASP after call site for BTI-enabled
kernel, so there is no problem. To fix it, replace BLR with RET to bypass
the branch target check.
Fixes: efc9909fdc ("bpf, arm64: Add bpf trampoline for arm64")
Reported-by: Florent Revest <revest@chromium.org>
Signed-off-by: Xu Kuohai <xukuohai@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Florent Revest <revest@chromium.org>
Acked-by: Florent Revest <revest@chromium.org>
Link: https://lore.kernel.org/bpf/20230401234144.3719742-1-xukuohai@huaweicloud.com
Add the IRQCHIP_SKIP_SET_WAKE flag since there are no special IRQ Wake
bits that can be set to enable wakeup IRQ.
Fixes: 3d9edf09d4 ("[ARM] 4457/2: davinci: GPIO support")
Signed-off-by: Dhruva Gole <d-gole@ti.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
The interrupt enable bits might be set if we want to use the GPIO as
wakeup source. Clearing this will mean disabling of interrupts in the GPIO
banks that we may want to wakeup from.
Thus remove the line that was clearing this bit from the driver's save
context function.
Cc: Devarsh Thakkar <devarsht@ti.com>
Fixes: 0651a73092 ("gpio: davinci: Add support for system suspend/resume PM")
Signed-off-by: Dhruva Gole <d-gole@ti.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Keerthy <j-keerthy@ti.com>
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Currently ath11k breaks after hibernation, the reason being that ath11k expects
that the wireless device will have power during suspend and the firmware will
continue running. But of course during hibernation the power from the device is
cut off and firmware is not running when resuming, so ath11k will fail.
(The reason why ath11k needs the firmware running is the interaction between
mac80211 and MHI stack, it's a long story and more info in the bugzilla report.)
In SUSE kernels the watchdog timeout is reduced from the default 120 to 60 seconds:
CONFIG_DPM_WATCHDOG_TIMEOUT=60
But as the ath11k MHI timeout is 90 seconds the kernel will crash before will
ath11k will recover in resume callback. To avoid the crash reduce the MHI
timeout to just 20 seconds.
Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.9
Link: https://bugzilla.kernel.org/show_bug.cgi?id=214649
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230329162038.8637-1-kvalo@kernel.org
Remove pci_clear_master to simplify the code,
the bus-mastering is also cleared in do_pci_disable_device,
like this:
./drivers/pci/pci.c:2197
static void do_pci_disable_device(struct pci_dev *dev)
{
u16 pci_command;
pci_read_config_word(dev, PCI_COMMAND, &pci_command);
if (pci_command & PCI_COMMAND_MASTER) {
pci_command &= ~PCI_COMMAND_MASTER;
pci_write_config_word(dev, PCI_COMMAND, pci_command);
}
pcibios_disable_device(dev);
}.
And dev->is_busmaster is set to 0 in pci_disable_device.
Signed-off-by: Cai Huoqing <cai.huoqing@linux.dev>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Acked-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230323112613.7550-5-cai.huoqing@linux.dev
We use ppdu_sts to obtain channel information from hardware, to ensure
that the scan results have correct channel information. However, some of
ppdu_sts that is filtered by the to_self check is also needed for the scan
results. Therefore, we modify the filter part in front of CFO count.
Signed-off-by: Chih-Kang Chang <gary.chang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230320093112.30466-3-pkshih@realtek.com
The register-based H2C/C2H are used to exchange information between driver
and firmware, but only apply to narrow area because its data size is
smaller than regular packet-based H2C/C2H.
This kind of H2C/C2H must be paired. To identify if any H2C/C2H is missing,
update counters to help diagnose this kind of problems.
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230316063956.71687-1-pkshih@realtek.com
On the remote side, when QRTR socket is removed, af_qrtr will call
qrtr_port_remove() which broadcasts the DEL_CLIENT packet to all neighbours
including local NS. NS upon receiving the DEL_CLIENT packet, will remove
the lookups associated with the node:port and broadcasts the DEL_SERVER
packet.
But on the host side, due to the arrival of the DEL_CLIENT packet, the NS
would've already deleted the server belonging to that port. So when the
remote's NS again broadcasts the DEL_SERVER for that port, it throws below
error message on the host:
"failed while handling packet from 2:-2"
So fix this error by not broadcasting the DEL_SERVER packet when the
DEL_CLIENT packet gets processed."
Fixes: 0c2204a4ad ("net: qrtr: Migrate nameservice to kernel from userspace")
Reviewed-by: Manivannan Sadhasivam <mani@kernel.org>
Signed-off-by: Ram Kumar Dharuman <quic_ramd@quicinc.com>
Signed-off-by: Sricharan Ramabadhran <quic_srichara@quicinc.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Golle says:
====================
net: dsa: add support for MT7988
The MediaTek MT7988 SoC comes with a built-in switch very similar to
previous MT7530 and MT7531. However, the switch address space is mapped
into the SoCs memory space rather than being connected via MDIO.
Using MMIO simplifies register access and also removes the need for a bus
lock, and for that reason also makes interrupt handling more light-weight.
Note that this is different from previous SoCs like MT7621 and MT7623N
which also came with an integrated MT7530-like switch which yet had to be
accessed via MDIO.
Split-off the part of the driver registering an MDIO driver, then add
another module acting as MMIO/platform driver.
The whole series has been tested on various MediaTek boards:
* MT7623A + MT7530 (BPi-R2)
* MT7986A + MT7531 (BPi-R3)
* MT7988A reference board
Changes since v1:
* use 'internal' PHY mode where appropriate
* use regmap_update_bits in mt7530_rmw
* improve dt-bindings
Changes since RFC v3:
* WARN_ON_ONCE if register read fails
* move probing of the reset GPIO and reset controller link out of
common probe function, as they are not actually common
Changes since RFC v2:
* split into many small commits to ease review
* introduce helper functions to reduce code duplication
* use helpers for locking to make lock-skipping easier and less ugly
to implement.
* add dt-bindings for mediatek,mt7988-switch
Changes since initial RFC:
* use regmap for register access and move register access to bus-
specific driver
* move initialization of MT7531 SGMII PCS to MDIO driver
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Add documentation for the built-in switch which can be found in the
MediaTek MT7988 SoC.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add driver for the built-in Gigabit Ethernet switch which can be found
in the MediaTek MT7988 SoC.
The switch shares most of its design with MT7530 and MT7531, but has
it's registers mapped into the SoCs register space rather than being
connected externally or internally via MDIO.
Introduce a new platform driver to support that.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
As MT7530 and MT7531 internally use 32-bit wide registers, each access
to any register of the switch requires several operations on the MDIO
bus. Hence if there is congruent access, e.g. due to PCS or PHY
polling, this can mess up and interfere with another ongoing register
access sequence.
However, the MDIO bus mutex is only relevant for MDIO-connected
switches. Prepare switches which have there registers directly mapped
into the SoCs register space via MMIO which do not require such
locking. There we can simply use regmap's default locking mechanism.
Hence guard mutex operations to only be performed in case of MDIO
connected switches.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Split MT7530 switch driver into a common part and a part specific
for MDIO connected switches and multi-chip modules.
Move MDIO-specific functions to newly introduced mt7530-mdio.c while
keeping the common parts in mt7530.c.
Introduce new Kconfig symbol CONFIG_NET_DSA_MT7530_MDIO which is
implied by CONFIG_NET_DSA_MT7530.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
MT7988 shares a significant part of the setup function with MT7531.
Split-off those parts into a shared function which is going to be used
also by mt7988_setup.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move commonly used parts from mt7530_remove into new
mt7530_remove_common helper function which will be used by both,
mt7530_remove and the to-be-introduced mt7988_remove.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move commonly used parts from mt7530_probe into new mt7530_probe_common
helper function which will be used by both, mt7530_probe and the
to-be-introduced mt7988_probe.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
In preparation of splitting mt7530.c into a driver for MDIO-connected
as well as MDIO-accessed built-in switches on one hand and MMIO-accessed
built-in switches move the p5_inft_modes() function from mt7530.h to
mt7530.c. The function is only needed there and will trigger a compiler
warning about a defined but unused function otherwise when including
mt7530.h in the to-be-introduced bus-specific drivers.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
As the MDIO bus lock only needs to be involved if actually operating
on an MDIO-connected switch we will need to skip locking for built-in
switches which are accessed via MMIO.
Create helper functions which simplify that upcoming change.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move creating the SGMII PCS from mt753x_setup() to the more appropriate
mt7530_probe() function.
This is done also in preparation of moving all functions related to
MDIO-connected MT753x switches to a separate module.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of wrapping the locked register accessor functions, use the
unlocked variants and add locking wrapper functions to let regmap
handle the locking.
This is a preparation towards being able to always use regmap to
access switch registers instead of open-coded accessor functions.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of macro templates use a dedidated function and allocated
regmap_config when creating the regmaps for the pcs-mtk-lynxi
instances.
This is in preparation to switching to use unlocked regmap accessors
and have regmap's locking API handle locking for us.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Simply returning the negative error value instead of the read value
doesn't seem like a good idea. Return 0 instead and add WARN_ON_ONCE(1)
so this kind of error will not go unnoticed.
Suggested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Heiner Kallweit says:
====================
net: phy: smsc: add support for edpd tunable
This adds support for the EDPD PHY tunable.
Per default EDPD is disabled in interrupt mode, the tunable can be used
to override this, e.g. if the link partner doesn't use EDPD.
The interval to check for energy can be chosen between 1000ms and
2000ms. Note that this value consists of the 1000ms phylib interval
for state machine runs plus the time to wait for energy being detected.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This adds support for the EDPD PHY tunable.
Per default EDPD is disabled in interrupt mode, the tunable can be used
to override this, e.g. if the link partner doesn't use EDPD.
The interval to check for energy can be chosen between 1000ms and
2000ms. Note that this value consists of the 1000ms phylib interval
for state machine runs plus the time to wait for energy being detected.
v2:
- consider that phylib core holds phydev->lock when calling the
phy tunable hooks
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a member edpd_max_wait_ms to the private data structure in preparation
of making the wait period configurable by supporting the edpd phy tunable.
v2:
- rename constant to EDPD_MAX_WAIT_DFLT_MS
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add flag edpd_mode_set_by_user in preparation of adding edpd phy tunable
support. This flag will allow users to override the default behavior
of edpd being disabled if interrupt mode is used.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Clear edpd_enable if interupt mode is used, this avoids
having to check for PHY_POLL multiple times.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
