In commit c8c3735997a3 ("parisc: Enhance detection of synchronous cr16
clocksources") I assumed that CPUs on the same physical core are syncronous.
While booting up the kernel on two different C8000 machines, one with a
dual-core PA8800 and one with a dual-core PA8900 CPU, this turned out to be
wrong. The symptom was that I saw a jump in the internal clocks printed to the
syslog and strange overall behaviour. On machines which have 4 cores (2
dual-cores) the problem isn't visible, because the current logic already marked
the cr16 clocksource unstable in this case.
This patch now marks the cr16 interval timers unstable if we have more than one
CPU in the system, and it fixes this issue.
Commit 761ed4a94582 ("tty: serial_core: convert uart_close to use
tty_port_close") converted serial core to use tty_port_close() but
failed to notice that the transmit buffer still needs to be freed on
final close.
Not freeing the transmit buffer means that the buffer is no longer
cleared on next open so that any ioctl() waiting for the buffer to drain
might wait indefinitely (e.g. on termios changes) or that stale data can
end up being transmitted in case tx is restarted.
Furthermore, the buffer of any port that has been opened would leak on
driver unbind.
Note that the port lock is held when clearing the buffer pointer due to
the ldisc race worked around by commit a5ba1d95e46e ("uart: fix race
between uart_put_char() and uart_shutdown()").
Also note that the tty-port shutdown() callback is not called for
console ports so it is not strictly necessary to free the buffer page
after releasing the lock (cf. d72402145ace ("tty/serial: do not free
trasnmit buffer page under port lock")).
The document 'ACPI for Arm Components 1.0' defines the following
_HID mappings:
-'Prime cell UART (PL011)': ARMH0011
-'SBSA UART': ARMHB000
Use the sbsa-uart driver when a device is described with
the 'ARMHB000' _HID.
Note:
PL011 devices currently use the sbsa-uart driver instead of the
uart-pl011 driver. Indeed, PL011 devices are not bound to a clock
in ACPI. It is not possible to change their baudrate.
The CONSOLE_POLLING mode is used for tools like k(g)db. In this kind of
setup, it is often sharing a serial device with the normal system console.
This is usually no problem because the polling helpers can consume input
values directly (when in kgdb context) and the normal Linux handlers can
only consume new input values after kgdb switched back.
This is not true anymore when RX DMA is enabled for UARTDM controllers.
Single input values can no longer be received correctly. Instead following
seems to happen:
* on 1. input, some old input is read (continuously)
* on 2. input, two old inputs are read (continuously)
* on 3. input, three old input values are read (continuously)
* on 4. input, 4 previous inputs are received
This repeats then for each group of 4 input values.
This behavior changes slightly depending on what state the controller was
when the first input was received. But this makes working with kgdb
basically impossible because control messages are always corrupted when
kgdboc tries to parse them.
RX DMA should therefore be off when CONSOLE_POLLING is enabled to avoid
these kind of problems. No such problem was noticed for TX DMA.
The trampoline_pgd only maps the 0xfffffff000000000-0xffffffffffffffff
range of kernel memory (with 4-level paging). This range contains the
kernel's text+data+bss mappings and the module mapping space but not the
direct mapping and the vmalloc area.
This is enough to get the application processors out of real-mode, but
for code that switches back to real-mode the trampoline_pgd is missing
important parts of the address space. For example, consider this code
from arch/x86/kernel/reboot.c, function machine_real_restart() for a
64-bit kernel:
The code switches to the trampoline_pgd, which unmaps the direct mapping
and also the kernel stack. The call to cr4_clear_bits() will find no
stack and crash the machine. The real_mode_header pointer below points
into the direct mapping, and dereferencing it also causes a crash.
The reason this does not crash always is only that kernel mappings are
global and the CR3 switch does not flush those mappings. But if theses
mappings are not in the TLB already, the above code will crash before it
can jump to the real-mode stub.
Extend the trampoline_pgd to contain all kernel mappings to prevent
these crashes and to make code which runs on this page-table more
robust.
Stub from the spec:
"4.5.2.2.4.2 Exiting from AttachWait.SNK State
A Sink shall transition to Unattached.SNK when the state of both
the CC1 and CC2 pins is SNK.Open for at least tPDDebounce.
A DRP shall transition to Unattached.SRC when the state of both
the CC1 and CC2 pins is SNK.Open for at least tPDDebounce."
This change makes TCPM to wait in SNK_DEBOUNCED state until
CC1 and CC2 pins is SNK.Open for at least tPDDebounce. Previously,
TCPM resets the port if vbus is not present in PD_T_PS_SOURCE_ON.
This causes TCPM to loop continuously when connected to a
faulty power source that does not present vbus. Waiting in
SNK_DEBOUNCED also ensures that TCPM is adherant to
"4.5.2.2.4.2 Exiting from AttachWait.SNK State" requirements.
Turns out some xHC controllers require all 64 bits in the CRCR register
to be written to execute a command abort.
The lower 32 bits containing the command abort bit is written first.
In case the command ring stops before we write the upper 32 bits then
hardware may use these upper bits to set the commnd ring dequeue pointer.
Solve this by making sure the upper 32 bits contain a valid command
ring dequeue pointer.
The original patch that only wrote the first 32 to stop the ring went
to stable, so this fix should go there as well.
Fixes: ff0e50d3564f ("xhci: Fix command ring pointer corruption while aborting a command") Cc: stable@vger.kernel.org Tested-by: Pavankumar Kondeti <quic_pkondeti@quicinc.com> Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com> Link: https://lore.kernel.org/r/20211126122340.1193239-2-mathias.nyman@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
caused by `c->vc_cell_height' not having been initialized. This has
only started to trigger with commit 860dafa90259 ("vt: Fix character
height handling with VT_RESIZEX"), however the ultimate offender is
commit 50ec42edd978 ("[PATCH] Detaching fbcon: fix vgacon to allow
retaking of the console").
Said commit has added a call to `vc_resize' whenever `vgacon_init' is
called with the `init' argument set to 0, which did not happen before.
And the call is made before a key vgacon boot parameter retrieved in
`vgacon_startup' has been propagated in `vgacon_init' for `vc_resize' to
use to the console structure being worked on. Previously the parameter
was `c->vc_font.height' and now it is `c->vc_cell_height'.
In this particular scenario the registration of fbcon has failed and vt
resorts to vgacon. Now fbcon does have initialized `c->vc_font.height'
somehow, unlike `c->vc_cell_height', which is why this code did not
crash before, but either way the boot parameters should have been copied
to the console structure ahead of the call to `vc_resize' rather than
afterwards, so that first the call has a chance to use them and second
they do not change the console structure to something possibly different
from what was used by `vc_resize'.
Move the propagation of the vgacon boot parameters ahead of the call to
`vc_resize' then. Adjust the comment accordingly.
Fixes: 50ec42edd978 ("[PATCH] Detaching fbcon: fix vgacon to allow retaking of the console") Cc: stable@vger.kernel.org # v2.6.18+ Reported-by: Wim Osterholt <wim@djo.tudelft.nl> Reported-by: Pavel V. Panteleev <panteleev_p@mcst.ru> Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk> Link: https://lore.kernel.org/r/alpine.DEB.2.21.2110252317110.58149@angie.orcam.me.uk Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On newer debian releases the debian-provided "installkernel" script is
installed in /usr/sbin. Fix the kernel install.sh script to look for the
script in this directory as well.
Default KBUILD_IMAGE to $(boot)/bzImage if a self-extracting
(CONFIG_PARISC_SELF_EXTRACT=y) kernel is to be built.
This fixes the bindeb-pkg make target.
In particular, we need to ensure all the necessary blocks are switched
to 64b mode (a5xx+) otherwise the high bits of the address of the BO to
snapshot state into will be ignored, resulting in:
Fixes: 4f776f4511c7 ("drm/msm/gpu: Convert the GPU show function to use the GPU state") Signed-off-by: Rob Clark <robdclark@chromium.org> Link: https://lore.kernel.org/r/20211108180122.487859-1-robdclark@gmail.com Signed-off-by: Rob Clark <robdclark@chromium.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When smc_close_final() returns error, the return code overwrites by
kernel_sock_shutdown() in smc_close_active(). The return code of
smc_close_final() is more important than kernel_sock_shutdown(), and it
will pass to userspace directly.
Fix it by keeping both return codes, if smc_close_final() raises an
error, return it or kernel_sock_shutdown()'s.
Link: https://lore.kernel.org/linux-s390/1f67548e-cbf6-0dce-82b5-10288a4583bd@linux.ibm.com/ Fixes: 606a63c9783a ("net/smc: Ensure the active closing peer first closes clcsock") Suggested-by: Karsten Graul <kgraul@linux.ibm.com> Signed-off-by: Tony Lu <tonylu@linux.alibaba.com> Reviewed-by: Wen Gu <guwen@linux.alibaba.com> Acked-by: Karsten Graul <kgraul@linux.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Correct an error where setting /proc/sys/net/rds/tcp/rds_tcp_rcvbuf would
instead modify the socket's sk_sndbuf and would leave sk_rcvbuf untouched.
Fixes: c6a58ffed536 ("RDS: TCP: Add sysctl tunables for sndbuf/rcvbuf on rds-tcp socket") Signed-off-by: William Kucharski <william.kucharski@oracle.com> Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reported by Kernel Concurrency Sanitizer on:
CPU: 1 PID: 28712 Comm: syz-executor.0 Tainted: G W 5.16.0-rc1-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Link: https://lore.kernel.org/r/20211130170155.2331929-1-eric.dumazet@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On most systems request for IRQ 0 will fail, phylib will print an error message
and fall back to polling. To fix this set the phydev->irq to PHY_POLL if no IRQ
is available.
Fixes: cc89c323a30e ("lan78xx: Use irq_domain for phy interrupt from USB Int. EP") Reviewed-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: Sven Schuchmann <schuchmann@schleissheimer.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Need to call rxrpc_put_local() for peer candidate before kfree() as it
holds a ref to rxrpc_local.
[DH: v2: Changed to abstract the peer freeing code out into a function]
Fixes: 9ebeddef58c4 ("rxrpc: rxrpc_peer needs to hold a ref on the rxrpc_local record") Signed-off-by: Eiichi Tsukata <eiichi.tsukata@nutanix.com> Signed-off-by: David Howells <dhowells@redhat.com> Reviewed-by: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org Link: https://lore.kernel.org/all/20211121041608.133740-2-eiichi.tsukata@nutanix.com/ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In mlx4_en_try_alloc_resources(), mlx4_en_copy_priv() is called and
tmp->tx_cq will be freed on the error path of mlx4_en_copy_priv().
After that mlx4_en_alloc_resources() is called and there is a dereference
of &tmp->tx_cq[t][i] in mlx4_en_alloc_resources(), which could lead to
a use after free problem on failure of mlx4_en_copy_priv().
Fix this bug by adding a check of mlx4_en_copy_priv()
This bug was found by a static analyzer. The analysis employs
differential checking to identify inconsistent security operations
(e.g., checks or kfrees) between two code paths and confirms that the
inconsistent operations are not recovered in the current function or
the callers, so they constitute bugs.
Note that, as a bug found by static analysis, it can be a false
positive or hard to trigger. Multiple researchers have cross-reviewed
the bug.
Builds with CONFIG_MLX4_EN=m show no new warnings,
and our static analyzer no longer warns about this code.
Fixes: ec25bc04ed8e ("net/mlx4_en: Add resilience in low memory systems") Signed-off-by: Zhou Qingyang <zhou1615@umn.edu> Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Link: https://lore.kernel.org/r/20211130164438.190591-1-zhou1615@umn.edu Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On ARM v6 and later, we define CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
because the ordinary load/store instructions (ldr, ldrh, ldrb) can
tolerate any misalignment of the memory address. However, load/store
double and load/store multiple instructions (ldrd, ldm) may still only
be used on memory addresses that are 32-bit aligned, and so we have to
use the CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS macro with care, or we
may end up with a severe performance hit due to alignment traps that
require fixups by the kernel. Testing shows that this currently happens
with clang-13 but not gcc-11. In theory, any compiler version can
produce this bug or other problems, as we are dealing with undefined
behavior in C99 even on architectures that support this in hardware,
see also https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100363.
Fortunately, the get_unaligned() accessors do the right thing: when
building for ARMv6 or later, the compiler will emit unaligned accesses
using the ordinary load/store instructions (but avoid the ones that
require 32-bit alignment). When building for older ARM, those accessors
will emit the appropriate sequence of ldrb/mov/orr instructions. And on
architectures that can truly tolerate any kind of misalignment, the
get_unaligned() accessors resolve to the leXX_to_cpup accessors that
operate on aligned addresses.
Since the compiler will in fact emit ldrd or ldm instructions when
building this code for ARM v6 or later, the solution is to use the
unaligned accessors unconditionally on architectures where this is
known to be fast. The _aligned version of the hash function is
however still needed to get the best performance on architectures
that cannot do any unaligned access in hardware.
This new version avoids the undefined behavior and should produce
the fastest hash on all architectures we support.
There are various problems related to netlink notifications for mpls route
changes in response to interfaces being deleted:
* delete interface of only nexthop
DELROUTE notification is missing RTA_OIF attribute
* delete interface of non-last nexthop
NEWROUTE notification is missing entirely
* delete interface of last nexthop
DELROUTE notification is missing nexthop
All of these problems stem from the fact that existing routes are modified
in-place before sending a notification. Restructure mpls_ifdown() to avoid
changing the route in the DELROUTE cases and to create a copy in the
NEWROUTE case.
Fixes: f8efb73c97e2 ("mpls: multipath route support") Signed-off-by: Benjamin Poirier <bpoirier@nvidia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In qlcnic_83xx_add_rings(), the indirect function of
ahw->hw_ops->alloc_mbx_args will be called to allocate memory for
cmd.req.arg, and there is a dereference of it in qlcnic_83xx_add_rings(),
which could lead to a NULL pointer dereference on failure of the
indirect function like qlcnic_83xx_alloc_mbx_args().
Fix this bug by adding a check of alloc_mbx_args(), this patch
imitates the logic of mbx_cmd()'s failure handling.
This bug was found by a static analyzer. The analysis employs
differential checking to identify inconsistent security operations
(e.g., checks or kfrees) between two code paths and confirms that the
inconsistent operations are not recovered in the current function or
the callers, so they constitute bugs.
Note that, as a bug found by static analysis, it can be a false
positive or hard to trigger. Multiple researchers have cross-reviewed
the bug.
Builds with CONFIG_QLCNIC=m show no new warnings, and our
static analyzer no longer warns about this code.
Fix section mismatch warnings in xtsonic. The first one appears to be
bogus and after fixing the second one, the first one is gone.
WARNING: modpost: vmlinux.o(.text+0x529adc): Section mismatch in reference from the function sonic_get_stats() to the function .init.text:set_reset_devices()
The function sonic_get_stats() references
the function __init set_reset_devices().
This is often because sonic_get_stats lacks a __init
annotation or the annotation of set_reset_devices is wrong.
WARNING: modpost: vmlinux.o(.text+0x529b3b): Section mismatch in reference from the function xtsonic_probe() to the function .init.text:sonic_probe1()
The function xtsonic_probe() references
the function __init sonic_probe1().
This is often because xtsonic_probe lacks a __init
annotation or the annotation of sonic_probe1 is wrong.
Fixes: 74f2a5f0ef64 ("xtensa: Add support for the Sonic Ethernet device for the XT2000 board.") Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Reported-by: kernel test robot <lkp@intel.com> Cc: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Cc: Finn Thain <fthain@telegraphics.com.au> Cc: Chris Zankel <chris@zankel.net> Cc: linux-xtensa@linux-xtensa.org Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Acked-by: Max Filippov <jcmvbkbc@gmail.com> Link: https://lore.kernel.org/r/20211130063947.7529-1-rdunlap@infradead.org Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In case of receiving a NACK, the dma transfer should be stopped
to avoid feeding data into the FIFO.
Also ensure to properly return the proper error code and avoid
waiting for the end of the dma completion in case of
error happening during the transmission.
Jann Horn points out that there is another possible race wrt Unix domain
socket garbage collection, somewhat reminiscent of the one fixed in
commit cbcf01128d0a ("af_unix: fix garbage collect vs MSG_PEEK").
See the extended comment about the garbage collection requirements added
to unix_peek_fds() by that commit for details.
The race comes from how we can locklessly look up a file descriptor just
as it is in the process of being closed, and with the right artificial
timing (Jann added a few strategic 'mdelay(500)' calls to do that), the
Unix domain socket garbage collector could see the reference count
decrement of the close() happen before fget() took its reference to the
file and the file was attached onto a new file descriptor.
This is all (intentionally) correct on the 'struct file *' side, with
RCU lookups and lockless reference counting very much part of the
design. Getting that reference count out of order isn't a problem per
se.
But the garbage collector can get confused by seeing this situation of
having seen a file not having any remaining external references and then
seeing it being attached to an fd.
In commit cbcf01128d0a ("af_unix: fix garbage collect vs MSG_PEEK") the
fix was to serialize the file descriptor install with the garbage
collector by taking and releasing the unix_gc_lock.
That's not really an option here, but since this all happens when we are
in the process of looking up a file descriptor, we can instead simply
just re-check that the file hasn't been closed in the meantime, and just
re-do the lookup if we raced with a concurrent close() of the same file
descriptor.
Some uses cases repeatedly get and put references to the same file, but
the only exposed interface is doing these one at the time. As each of
these entail an atomic inc or dec on a shared structure, that cost can
add up.
Add fget_many(), which works just like fget(), except it takes an
argument for how many references to get on the file. Ditto fput_many(),
which can drop an arbitrary number of references to a file.
Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Trying to remove the fsl-sata module in the PPC64 GNU/Linux
leads to the following warning:
------------[ cut here ]------------
remove_proc_entry: removing non-empty directory 'irq/69',
leaking at least 'fsl-sata[ff0221000.sata]'
WARNING: CPU: 3 PID: 1048 at fs/proc/generic.c:722
.remove_proc_entry+0x20c/0x220
IRQMASK: 0
NIP [c00000000033826c] .remove_proc_entry+0x20c/0x220
LR [c000000000338268] .remove_proc_entry+0x208/0x220
Call Trace:
.remove_proc_entry+0x208/0x220 (unreliable)
.unregister_irq_proc+0x104/0x140
.free_desc+0x44/0xb0
.irq_free_descs+0x9c/0xf0
.irq_dispose_mapping+0x64/0xa0
.sata_fsl_remove+0x58/0xa0 [sata_fsl]
.platform_drv_remove+0x40/0x90
.device_release_driver_internal+0x160/0x2c0
.driver_detach+0x64/0xd0
.bus_remove_driver+0x70/0xf0
.driver_unregister+0x38/0x80
.platform_driver_unregister+0x14/0x30
.fsl_sata_driver_exit+0x18/0xa20 [sata_fsl]
---[ end trace 0ea876d4076908f5 ]---
The driver creates the mapping by calling irq_of_parse_and_map(),
so it also has to dispose the mapping. But the easy way out is to
simply use platform_get_irq() instead of irq_of_parse_map(). Also
we should adapt return value checking and propagate error values.
In this case the mapping is not managed by the device but by
the of core, so the device has not to dispose the mapping.
Fixes: faf0b2e5afe7 ("drivers/ata: add support to Freescale 3.0Gbps SATA Controller") Cc: stable@vger.kernel.org Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Baokun Li <libaokun1@huawei.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@gmail.com> Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the `rmmod sata_fsl.ko` command is executed in the PPC64 GNU/Linux,
a bug is reported:
==================================================================
BUG: Unable to handle kernel data access on read at 0x80000800805b502c
Oops: Kernel access of bad area, sig: 11 [#1]
NIP [c0000000000388a4] .ioread32+0x4/0x20
LR [80000000000c6034] .sata_fsl_port_stop+0x44/0xe0 [sata_fsl]
Call Trace:
.free_irq+0x1c/0x4e0 (unreliable)
.ata_host_stop+0x74/0xd0 [libata]
.release_nodes+0x330/0x3f0
.device_release_driver_internal+0x178/0x2c0
.driver_detach+0x64/0xd0
.bus_remove_driver+0x70/0xf0
.driver_unregister+0x38/0x80
.platform_driver_unregister+0x14/0x30
.fsl_sata_driver_exit+0x18/0xa20 [sata_fsl]
.__se_sys_delete_module+0x1ec/0x2d0
.system_call_exception+0xfc/0x1f0
system_call_common+0xf8/0x200
==================================================================
The triggering of the BUG is shown in the following stack:
The iounmap(host_priv->hcr_base) and kfree(host_priv) functions should
not be executed in drv->remove. These functions should be executed in
host_stop after port_stop. Therefore, we move these functions to the
new function sata_fsl_host_stop and bind the new function to host_stop.
Fixes: faf0b2e5afe7 ("drivers/ata: add support to Freescale 3.0Gbps SATA Controller") Cc: stable@vger.kernel.org Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Baokun Li <libaokun1@huawei.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@gmail.com> Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently when removing an ipmi_user the removal is deferred as a work on
the system's workqueue. Although this guarantees the free operation will
occur in non atomic context, it can race with the ipmi_msghandler module
removal (see [1]) . In case a remove_user work is scheduled for removal
and shortly after ipmi_msghandler module is removed we can end up in a
situation where the module is removed fist and when the work is executed
the system crashes with :
BUG: unable to handle page fault for address: ffffffffc05c3450
PF: supervisor instruction fetch in kernel mode
PF: error_code(0x0010) - not-present page
because the pages of the module are gone. In cleanup_ipmi() there is no
easy way to detect if there are any pending works to flush them before
removing the module. This patch creates a separate workqueue and schedules
the remove_work works on it. When removing the module the workqueue is
drained when destroyed to avoid the race.
The 'kprobe::data_size' is unsigned, thus it can not be negative. But if
user sets it enough big number (e.g. (size_t)-8), the result of 'data_size
+ sizeof(struct kretprobe_instance)' becomes smaller than sizeof(struct
kretprobe_instance) or zero. In result, the kretprobe_instance are
allocated without enough memory, and kretprobe accesses outside of
allocated memory.
To avoid this issue, introduce a max limitation of the
kretprobe::data_size. 4KB per instance should be OK.
IPCB/IP6CB need to be initialized when processing outbound v4 or v6 pkts
in the codepath of vrf device xmit function so that leftover garbage
doesn't cause futher code that uses the CB to incorrectly process the
pkt.
One occasion of the issue might occur when MPLS route uses the vrf
device as the outgoing device such as when the route is added using "ip
-f mpls route add <label> dev <vrf>" command.
The problems seems to exist since day one. Hence I put the day one
commits on the Fixes tags.
Fixes: 193125dbd8eb ("net: Introduce VRF device driver") Fixes: 35402e313663 ("net: Add IPv6 support to VRF device") Cc: stable@vger.kernel.org Signed-off-by: Stephen Suryaputra <ssuryaextr@gmail.com> Reviewed-by: David Ahern <dsahern@kernel.org> Link: https://lore.kernel.org/r/20211130162637.3249-1-ssuryaextr@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The definition of macro MOTO_SROM_BUG is:
#define MOTO_SROM_BUG (lp->active == 8 && (get_unaligned_le32(
dev->dev_addr) & 0x00ffffff) == 0x3e0008)
and the if statement
if (MOTO_SROM_BUG) lp->active = 0;
using this macro indicates lp->active could be 8. If lp->active is 8 and
the second comparison of this macro is false. lp->active will remain 8 in:
lp->phy[lp->active].gep = (*p ? p : NULL); p += (2 * (*p) + 1);
lp->phy[lp->active].rst = (*p ? p : NULL); p += (2 * (*p) + 1);
lp->phy[lp->active].mc = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].ana = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].fdx = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].ttm = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].mci = *p;
However, the length of array lp->phy is 8, so array overflows can occur.
To fix these possible array overflows, we first check lp->active and then
return -EINVAL if it is greater or equal to ARRAY_SIZE(lp->phy) (i.e. 8).
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn> Signed-off-by: Teng Qi <starmiku1207184332@gmail.com> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The if statement:
if (port >= DSAF_GE_NUM)
return;
limits the value of port less than DSAF_GE_NUM (i.e., 8).
However, if the value of port is 6 or 7, an array overflow could occur:
port_rst_off = dsaf_dev->mac_cb[port]->port_rst_off;
because the length of dsaf_dev->mac_cb is DSAF_MAX_PORT_NUM (i.e., 6).
To fix this possible array overflow, we first check port and if it is
greater than or equal to DSAF_MAX_PORT_NUM, the function returns.
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn> Signed-off-by: Teng Qi <starmiku1207184332@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
AMD requires that the SATA controller be configured for devsleep in order
for S0i3 entry to work properly.
commit b1a9585cc396 ("ata: ahci: Enable DEVSLP by default on x86 with
SLP_S0") sets up a kernel policy to enable devsleep on Intel mobile
platforms that are using s0ix. Add the PCI ID for the SATA controller in
Green Sardine platforms to extend this policy by default for AMD based
systems using s0i3 as well.
Cc: Nehal-bakulchandra Shah <Nehal-bakulchandra.Shah@amd.com> BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=214091 Signed-off-by: Mario Limonciello <mario.limonciello@amd.com> Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
We can race where iscsi_session_recovery_timedout() has woken up the error
handler thread and it's now setting the devices to offline, and
session_recovery_timedout()'s call to scsi_target_unblock() is also trying
to set the device's state to transport-offline. We can then get a mix of
states.
For the case where we can't relogin we want the devices to be in
transport-offline so when we have repaired the connection
__iscsi_unblock_session() can set the state back to running.
Set the device state then call into libiscsi to wake up the error handler.
During the suspend is in process, thermal_zone_device_update bails out
thermal zone re-evaluation for any sensor trip violation without
setting next valid trip to that sensor. It assumes during resume
it will re-evaluate same thermal zone and update trip. But when it is
in suspend temperature goes down and on resume path while updating
thermal zone if temperature is less than previously violated trip,
thermal zone set trip function evaluates the same previous high and
previous low trip as new high and low trip. Since there is no change
in high/low trip, it bails out from thermal zone set trip API without
setting any trip. It leads to a case where sensor high trip or low
trip is disabled forever even though thermal zone has a valid high
or low trip.
During thermal zone device init, reset thermal zone previous high
and low trip. It resolves above mentioned scenario.
Signed-off-by: Manaf Meethalavalappu Pallikunhi <manafm@codeaurora.org> Reviewed-by: Thara Gopinath <thara.gopinath@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When a disk has write caching disabled, we skip submission of a bio with
flush and sync requests before writing the superblock, since it's not
needed. However when the integrity checker is enabled, this results in
reports that there are metadata blocks referred by a superblock that
were not properly flushed. So don't skip the bio submission only when
the integrity checker is enabled for the sake of simplicity, since this
is a debug tool and not meant for use in non-debug builds.
fstests/btrfs/220 trigger a check-integrity warning like the following
when CONFIG_BTRFS_FS_CHECK_INTEGRITY=y and the disk with WCE=0.
There is a difference in how architectures treat "mem=" option. For some
that is an amount of online memory, for s390 and x86 this is the limiting
max address. Some memblock api like memblock_enforce_memory_limit()
take limit argument and explicitly treat it as the size of online memory,
and use __find_max_addr to convert it to an actual max address. Current
s390 usage:
yields different results depending on presence of memory holes (offline
memory blocks in between online memory). If there are no memory holes
limit == max_addr in memblock_enforce_memory_limit() and it does trim
online memory and reserved memory regions. With memory holes present it
actually does nothing.
Since we already use memblock_remove() explicitly to trim online memory
regions to potential limit (think mem=, kdump, addressing limits, etc.)
drop the usage of memblock_enforce_memory_limit() altogether. Trimming
reserved regions should not be required, since we now use
memblock_set_current_limit() to limit allocations and any explicit memory
reservations above the limit is an actual problem we should not hide.
When WWAN device wake from S3 deep, under thinkpad platform,
WWAN would be disabled. This disable status could be checked
by command 'nmcli r wwan' or 'rfkill list'.
Issue analysis as below:
When host resume from S3 deep, thinkpad_acpi driver would
call hotkey_resume() function. Finnaly, it will use
wan_get_status to check the current status of WWAN device.
During this resume progress, wan_get_status would always
return off even WWAN boot up completely.
In patch V2, Hans said 'sw_state should be unchanged
after a suspend/resume. It's better to drop the
tpacpi_rfk_update_swstate call all together from the
resume path'.
And it's confimed by Lenovo that GWAN is no longer
available from WHL generation because the design does not
match with current pin control.
Signed-off-by: Slark Xiao <slark_xiao@163.com> Link: https://lore.kernel.org/r/20211108060648.8212-1-slark_xiao@163.com Reviewed-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When kmemdup called failed and register_net_sysctl return NULL, should
return ENOMEM instead of ENOBUFS
Signed-off-by: liuguoqiang <liuguoqiang@uniontech.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
This bug report shows up when running our research tools. The
reports is SOOB read, but it seems SOOB write is also possible
a few lines below.
In details, fw.len and sw.len are inputs coming from io. A len
over the size of self->rpc triggers SOOB. The patch fixes the
bugs by adding sanity checks.
The bugs are triggerable with compromised/malfunctioning devices.
They are potentially exploitable given they first leak up to
0xffff bytes and able to overwrite the region later.
The patch is tested with QEMU emulater.
This is NOT tested with a real device.
Fix the length of holes reported at the end of a file: the length is
relative to the beginning of the extent, not the seek position which is
rounded down to the filesystem block size.
This bug went unnoticed for some time, but is now caught by the
following assertion in iomap_iter_done():
include/linux/of_clk.h:11:45: warning: ‘struct device_node’ declared inside parameter list will not be visible outside of this definition or declaration
unsigned int of_clk_get_parent_count(struct device_node *np);
^~~~~~~~~~~
include/linux/of_clk.h:12:43: warning: ‘struct device_node’ declared inside parameter list will not be visible outside of this definition or declaration
const char *of_clk_get_parent_name(struct device_node *np, int index);
^~~~~~~~~~~
include/linux/of_clk.h:13:31: warning: ‘struct of_device_id’ declared inside parameter list will not be visible outside of this definition or declaration
void of_clk_init(const struct of_device_id *matches);
^~~~~~~~~~~~
Fix this by adding forward declarations for struct device_node and
struct of_device_id.
The mechanism in use to allow the client to see the results of COPY/CLONE
is to drop those pages from the pagecache. This forces the client to read
those pages once more from the server. However, truncate_pagecache_range()
zeros out partial pages instead of dropping them. Let us instead use
invalidate_inode_pages2_range() with full-page offsets to ensure the client
properly sees the results of COPY/CLONE operations.
Currently, the exit_shm() function not designed to work properly when
task->sysvshm.shm_clist holds shm objects from different IPC namespaces.
This is a real pain when sysctl kernel.shm_rmid_forced = 1, because it
leads to use-after-free (reproducer exists).
This is an attempt to fix the problem by extending exit_shm mechanism to
handle shm's destroy from several IPC ns'es.
To achieve that we do several things:
1. add a namespace (non-refcounted) pointer to the struct shmid_kernel
2. during new shm object creation (newseg()/shmget syscall) we
initialize this pointer by current task IPC ns
3. exit_shm() fully reworked such that it traverses over all shp's in
task->sysvshm.shm_clist and gets IPC namespace not from current task
as it was before but from shp's object itself, then call
shm_destroy(shp, ns).
Note: We need to be really careful here, because as it was said before
(1), our pointer to IPC ns non-refcnt'ed. To be on the safe side we
using special helper get_ipc_ns_not_zero() which allows to get IPC ns
refcounter only if IPC ns not in the "state of destruction".
Q/A
Q: Why can we access shp->ns memory using non-refcounted pointer?
A: Because shp object lifetime is always shorther than IPC namespace
lifetime, so, if we get shp object from the task->sysvshm.shm_clist
while holding task_lock(task) nobody can steal our namespace.
Q: Does this patch change semantics of unshare/setns/clone syscalls?
A: No. It's just fixes non-covered case when process may leave IPC
namespace without getting task->sysvshm.shm_clist list cleaned up.
Link: https://lkml.kernel.org/r/67bb03e5-f79c-1815-e2bf-949c67047418@colorfullife.com Link: https://lkml.kernel.org/r/20211109151501.4921-1-manfred@colorfullife.com Fixes: ab602f79915 ("shm: make exit_shm work proportional to task activity") Co-developed-by: Manfred Spraul <manfred@colorfullife.com> Signed-off-by: Manfred Spraul <manfred@colorfullife.com> Signed-off-by: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: Greg KH <gregkh@linuxfoundation.org> Cc: Andrei Vagin <avagin@gmail.com> Cc: Pavel Tikhomirov <ptikhomirov@virtuozzo.com> Cc: Vasily Averin <vvs@virtuozzo.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Xen frontends shouldn't BUG() in case of illegal data received from
their backends. So replace the BUG_ON()s when reading illegal data from
the ring page with negative return values.
Today netfront will trust the backend to send only sane response data.
In order to avoid privilege escalations or crashes in case of malicious
backends verify the data to be within expected limits. Especially make
sure that the response always references an outstanding request.
Note that only the tx queue needs special id handling, as for the rx
queue the id is equal to the index in the ring page.
Introduce a new indicator for the device whether it is broken and let
the device stop working when it is set. Set this indicator in case the
backend sets any weird data.
Signed-off-by: Juergen Gross <jgross@suse.com> Reviewed-by: Jan Beulich <jbeulich@suse.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The tx_skb_freelist elements are in a single linked list with the
request id used as link reference. The per element link field is in a
union with the skb pointer of an in use request.
Move the link reference out of the union in order to enable a later
reuse of it for requests which need a populated skb pointer.
Rename add_id_to_freelist() and get_id_from_freelist() to
add_id_to_list() and get_id_from_list() in order to prepare using
those for other lists as well. Define ~0 as value to indicate the end
of a list and place that value into the link for a request not being
on the list.
When freeing a skb zero the skb pointer in the request. Use a NULL
value of the skb pointer instead of skb_entry_is_link() for deciding
whether a request has a skb linked to it.
Remove skb_entry_set_link() and open code it instead as it is really
trivial now.
Signed-off-by: Juergen Gross <jgross@suse.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In order to avoid a malicious backend being able to influence the local
processing of a request build the request locally first and then copy
it to the ring page. Any reading from the request influencing the
processing in the frontend needs to be done on the local instance.
Signed-off-by: Juergen Gross <jgross@suse.com> Reviewed-by: Jan Beulich <jbeulich@suse.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In order to avoid problems in case the backend is modifying a response
on the ring page while the frontend has already seen it, just read the
response into a local buffer in one go and then operate on that buffer
only.
Signed-off-by: Juergen Gross <jgross@suse.com> Reviewed-by: Jan Beulich <jbeulich@suse.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Today blkfront will trust the backend to send only sane response data.
In order to avoid privilege escalations or crashes in case of malicious
backends verify the data to be within expected limits. Especially make
sure that the response always references an outstanding request.
Introduce a new state of the ring BLKIF_STATE_ERROR which will be
switched to in case an inconsistency is being detected. Recovering from
this state is possible only via removing and adding the virtual device
again (e.g. via a suspend/resume cycle).
Make all warning messages issued due to valid error responses rate
limited in order to avoid message floods being triggered by a malicious
backend.
In order to avoid a malicious backend being able to influence the local
copy of a request build the request locally first and then copy it to
the ring page instead of doing it the other way round as today.
In order to avoid problems in case the backend is modifying a response
on the ring page while the frontend has already seen it, just read the
response into a local buffer in one go and then operate on that buffer
only.
Checking buf->flags should be done before the pipe_buf_release() is called
on the pipe buffer, since releasing the buffer might modify the flags.
This is exactly what page_cache_pipe_buf_release() does, and which results
in the same VM_BUG_ON_PAGE(PageLRU(page)) that the original patch was
trying to fix.
For the above two UAF, the root cause is that the nfc_dev_up can race
between the nci_unregister_device routine. Therefore, this patch
introduce NCI_UNREG flag to easily eliminate the possible race. In
addition, the mutex_lock in nci_close_device can act as a barrier.
Signed-off-by: Lin Ma <linma@zju.edu.cn> Fixes: 6a2968aaf50c ("NFC: basic NCI protocol implementation") Reviewed-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> Link: https://lore.kernel.org/r/20211116152732.19238-1-linma@zju.edu.cn Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When __unmap_hugepage_range() calls to huge_pmd_unshare() succeed, a TLB
flush is missing. This TLB flush must be performed before releasing the
i_mmap_rwsem, in order to prevent an unshared PMDs page from being
released and reused before the TLB flush took place.
Arguably, a comprehensive solution would use mmu_gather interface to
batch the TLB flushes and the PMDs page release, however it is not an
easy solution: (1) try_to_unmap_one() and try_to_migrate_one() also call
huge_pmd_unshare() and they cannot use the mmu_gather interface; and (2)
deferring the release of the page reference for the PMDs page until
after i_mmap_rwsem is dropeed can confuse huge_pmd_unshare() into
thinking PMDs are shared when they are not.
Fix __unmap_hugepage_range() by adding the missing TLB flush, and
forcing a flush when unshare is successful.
Fixes: 24669e58477e ("hugetlb: use mmu_gather instead of a temporary linked list for accumulating pages)" # 3.6 Signed-off-by: Nadav Amit <namit@vmware.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We should not walk/touch page tables outside of VMA boundaries when
holding only the mmap sem in read mode. Evil user space can modify the
VMA layout just before this function runs and e.g., trigger races with
page table removal code since commit dd2283f2605e ("mm: mmap: zap pages
with read mmap_sem in munmap"). gfn_to_hva() will only translate using
KVM memory regions, but won't validate the VMA.
Further, we should not allocate page tables outside of VMA boundaries: if
evil user space decides to map hugetlbfs to these ranges, bad things will
happen because we suddenly have PTE or PMD page tables where we
shouldn't have them.
Similarly, we have to check if we suddenly find a hugetlbfs VMA, before
calling get_locked_pte().
When pid filtering is activated in an instance, all of the events trace
files for that instance has the PID_FILTER flag set. This determines
whether or not pid filtering needs to be done on the event, otherwise the
event is executed as normal.
If pid filtering is enabled when an event is created (via a dynamic event
or modules), its flag is not updated to reflect the current state, and the
events are not filtered properly.
The "used length" reported by calling vhost_add_used() must be the
number of bytes written by the device (using "in" buffers).
In vhost_vsock_handle_tx_kick() the device only reads the guest
buffers (they are all "out" buffers), without writing anything,
so we must pass 0 as "used length" to comply virtio spec.
Fixes: 433fc58e6bf2 ("VSOCK: Introduce vhost_vsock.ko") Cc: stable@vger.kernel.org Reported-by: Halil Pasic <pasic@linux.ibm.com> Suggested-by: Jason Wang <jasowang@redhat.com> Signed-off-by: Stefano Garzarella <sgarzare@redhat.com> Link: https://lore.kernel.org/r/20211122163525.294024-2-sgarzare@redhat.com Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Halil Pasic <pasic@linux.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When PF is set to multi-TCs and configured mapping relationship between
priorities and TCs, the hardware will active these settings for this PF
and its VFs.
In this case when VF just uses one TC and its rx packets contain priority,
and if the priority is not mapped to TC0, as other TCs of VF is not valid,
hardware always put this kind of packets to the queue 0. It cause this kind
of packets of VF can not be used RSS function.
To fix this problem, set tc mode of all unused TCs of VF to the setting of
TC0, then rx packet with priority which map to unused TC will be direct to
TC0.
When applications call shutdown() with SHUT_RDWR in userspace,
smc_close_active() calls kernel_sock_shutdown(), and it is called
twice in smc_shutdown().
This fixes this by checking sk_state before do clcsock shutdown, and
avoids missing the application's call of smc_shutdown().
It hangup when booting Loongson 3A1000 with BOTH
CONFIG_PAGE_SIZE_64KB and CONFIG_MIPS_VA_BITS_48, that it turn
out to use 2-level pgtable instead of 3-level. 64KB page size
with 2-level pgtable only cover 42 bits VA, use 3-level pgtable
to cover all 48 bits VA(55 bits)
Fixes: 1e321fa917fb ("MIPS64: Support of at least 48 bits of SEGBITS) Signed-off-by: Huang Pei <huangpei@loongson.cn> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
While testing BIG TCP patch series, I was expecting that TCP_RR workloads
with 80KB requests/answers would send one 80KB TSO packet,
then being received as a single GRO packet.
It turns out this was not happening, and the root cause was that
cubic Hystart ACK train was triggering after a few (2 or 3) rounds of RPC.
Hystart was wrongly setting CWND/SSTHRESH to 30, while my RPC
needed a budget of ~20 segments.
Ideally these TCP_RR flows should not exit slow start.
Cubic Hystart should reset itself at each round, instead of assuming
every TCP flow is a bulk one.
Note that even after this patch, Hystart can still trigger, depending
on scheduling artifacts, but at a higher CWND/SSTHRESH threshold,
keeping optimal TSO packet sizes.
Tested:
ip link set dev eth0 gro_ipv6_max_size 131072 gso_ipv6_max_size 131072
nstat -n; netperf -H ... -t TCP_RR -l 5 -- -r 80000,80000 -K cubic; nstat|egrep "Ip6InReceives|Hystart|Ip6OutRequests"
Commit 39fbef4b0f77 ("PM: hibernate: Get block device exclusively in
swsusp_check()") changed the opening mode of the block device to
(FMODE_READ | FMODE_EXCL).
In the corresponding calls to swsusp_close(), the mode is still just
FMODE_READ which triggers the warning in blkdev_flush_mapping() on
resume from hibernate.
So, use the mode (FMODE_READ | FMODE_EXCL) also when closing the
device.
Fixes: 39fbef4b0f77 ("PM: hibernate: Get block device exclusively in swsusp_check()") Signed-off-by: Thomas Zeitlhofer <thomas.zeitlhofer+lkml@ze-it.at> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The side that actively closed socket, it's clcsock doesn't enter
TIME_WAIT state, but the passive side does it. It should show the same
behavior as TCP sockets.
Consider this, when client actively closes the socket, the clcsock in
server enters TIME_WAIT state, which means the address is occupied and
won't be reused before TIME_WAIT dismissing. If we restarted server, the
service would be unavailable for a long time.
To solve this issue, shutdown the clcsock in [A], perform the TCP active
close progress first, before the passive closed side closing it. So that
the actively closed side enters TIME_WAIT, not the passive one.
Client | Server
close() // client actively close |
smc_release() |
smc_close_active() // PEERCLOSEWAIT1 |
smc_close_final() // abort or closed = 1|
smc_cdc_get_slot_and_msg_send() |
[A] |
|smc_cdc_msg_recv_action() // ACTIVE
| queue_work(smc_close_wq, &conn->close_work)
| smc_close_passive_work() // PROCESSABORT or APPCLOSEWAIT1
| smc_close_passive_abort_received() // only in abort
|
|close() // server recv zero, close
| smc_release() // PROCESSABORT or APPCLOSEWAIT1
| smc_close_active()
| smc_close_abort() or smc_close_final() // CLOSED
| smc_cdc_get_slot_and_msg_send() // abort or closed = 1
smc_cdc_msg_recv_action() | smc_clcsock_release()
queue_work(smc_close_wq, &conn->close_work) | sock_release(tcp) // actively close clc, enter TIME_WAIT
smc_close_passive_work() // PEERCLOSEWAIT1 | smc_conn_free()
smc_close_passive_abort_received() // CLOSED|
smc_conn_free() |
smc_clcsock_release() |
sock_release(tcp) // passive close clc |
Link: https://www.spinics.net/lists/netdev/msg780407.html Fixes: b38d732477e4 ("smc: socket closing and linkgroup cleanup") Signed-off-by: Tony Lu <tonylu@linux.alibaba.com> Reviewed-by: Wen Gu <guwen@linux.alibaba.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The ->gem_create_object() functions are supposed to return NULL if there
is an error. None of the callers expect error pointers so returing one
will lead to an Oops. See drm_gem_vram_create(), for example.
While looping over shost's sdev list it is possible that one
of the drives is getting removed and its sas_target object is
freed but its sdev object remains intact.
Consequently, a kernel panic can occur while the driver is trying to access
the sas_address field of sas_target object without also checking the
sas_target object for NULL.
Link: https://lore.kernel.org/r/20211117104909.2069-1-sreekanth.reddy@broadcom.com Fixes: f92363d12359 ("[SCSI] mpt3sas: add new driver supporting 12GB SAS") Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When CONFIG_FORTIFY_SOURCE is set, memcpy() checks the potential
buffer overflow and panics. The code in sofcpga bootstrapping
contains the memcpy() calls are mistakenly translated as the shorter
size, hence it triggers a panic as if it were overflowing.
This patch changes the secondary_trampoline and *_end definitions
to arrays for avoiding the false-positive crash above.
The failure to retrieve post-op attributes has no bearing on whether or
not the clone operation itself was successful. We must therefore ignore
the return value of decode_getfattr() when looking at the success or
failure of nfs4_xdr_dec_clone().
This patch fixes an issue that an u32 netlink value is handled as a
signed enum value which doesn't fit into the range of u32 netlink type.
If it's handled as -1 value some BIT() evaluation ends in a
shift-out-of-bounds issue. To solve the issue we set the to u32 max which
is s32 "-1" value to keep backwards compatibility and let the followed enum
values start counting at 0. This brings the compiler to never handle the
enum as signed and a check if the value is above NL802154_IFTYPE_MAX should
filter -1 out.
Fixes: f3ea5e44231a ("ieee802154: add new interface command") Signed-off-by: Alexander Aring <aahringo@redhat.com> Link: https://lore.kernel.org/r/20211112030916.685793-1-aahringo@redhat.com Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
snd_ctl_remove() has to be called with card->controls_rwsem held (when
called after the card instantiation). This patch add the missing
rwsem calls around it.
would set the MultiMedia1 steam for SLIMBUS_0_RX, however doing below
command will reset previously setup MultiMedia1 stream, because both of them
are using MultiMedia1 PCM stream.
The GPIO controller is also an interrupt controller provider and is
currently missing the appropriate 'interrupt-controller' and
'#interrupt-cells' properties to denote that.
Fixes: fb026d3de33b ("ARM: BCM5301X: Add Broadcom's bus-axi to the DTS file") Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The I2C interrupt controller line is off by 32 because the datasheet
describes interrupt inputs into the GIC which are for Shared Peripheral
Interrupts and are starting at offset 32. The ARM GIC binding expects
the SPI interrupts to be numbered from 0 relative to the SPI base.
Fixes: bb097e3e0045 ("ARM: dts: BCM5301X: Add I2C support to the DT") Tested-by: Christian Lamparter <chunkeey@gmail.com> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
We are changing expire_nodest_conn to work even for reused connections when
conn_reuse_mode=0, just as what was done with commit dc7b3eb900aa ("ipvs:
Fix reuse connection if real server is dead").
For controlled and persistent connections, the new connection will get the
needed real server depending on the rules in ip_vs_check_template().
Fixes: d752c3645717 ("ipvs: allow rescheduling of new connections when port reuse is detected") Co-developed-by: Chuanqi Liu <legend050709@qq.com> Signed-off-by: Chuanqi Liu <legend050709@qq.com> Signed-off-by: yangxingwu <xingwu.yang@gmail.com> Acked-by: Simon Horman <horms@verge.net.au> Acked-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
We found out that we are unable to control the PERST# signal via the
default pin dedicated to be PERST# pin (GPIO2[3] pin) on A3700 SOC when
this pin is in EP_PCIE1_Resetn mode. There is a register in the PCIe
register space called PERSTN_GPIO_EN (D0088004[3]), but changing the
value of this register does not change the pin output when measuring
with voltmeter.
We do not know if this is a bug in the SOC, or if it works only when
PCIe controller is in a certain state.
Commit f4c7d053d7f7 ("PCI: aardvark: Wait for endpoint to be ready
before training link") says that when this pin changes pinctrl mode
from EP_PCIE1_Resetn to GPIO, the PERST# signal is asserted for a brief
moment.
So currently the situation is that on A3700 boards the PERST# signal is
asserted in U-Boot (because the code in U-Boot issues reset via this pin
via GPIO mode), and then in Linux by the obscure and undocumented
mechanism described by the above mentioned commit.
We want to issue PERST# signal in a known way, therefore this patch
changes the pcie_reset_pin function from "pcie" to "gpio" and adds the
reset-gpios property to the PCIe node in device tree files of
EspressoBin and Armada 3720 Dev Board (Turris Mox device tree already
has this property and uDPU does not have a PCIe port).
Signed-off-by: Marek Behún <marek.behun@nic.cz> Cc: Remi Pommarel <repk@triplefau.lt> Tested-by: Tomasz Maciej Nowak <tmn505@gmail.com> Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com> Signed-off-by: Gregory CLEMENT <gregory.clement@bootlin.com> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The PWM pins on North Bridge on Armada 37xx can be configured into PWM
or GPIO functions. When in PWM function, each pin can also be configured
to drive low on 0 and tri-state on 1 (LED mode).
The current definitions handle this by declaring two pin groups for each
pin:
- group "pwmN" with functions "pwm" and "gpio"
- group "ledN_od" ("od" for open drain) with functions "led" and "gpio"
This is semantically incorrect. The correct definition for each pin
should be one group with three functions: "pwm", "led" and "gpio".
Change the "pwmN" groups to support "led" function.
Remove "ledN_od" groups. This cannot break backwards compatibility with
older device trees: no device tree uses it since there is no PWM driver
for this SOC yet. Also "ledN_od" groups are not even documented.
Fixes: b835d6953009 ("pinctrl: armada-37xx: swap polarity on LED group") Signed-off-by: Marek Behún <kabel@kernel.org> Acked-by: Rob Herring <robh@kernel.org> Link: https://lore.kernel.org/r/20210719112938.27594-1-kabel@kernel.org Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This is a cleanup and fix of the patch by Ken Ma <make@marvell.com>.
Fix the mpp definitions according to newest revision of the
specification:
- northbridge:
fix pmic1 gpio number to 7
fix pmic0 gpio number to 6
- southbridge
split pcie1 group bit mask to BIT(5) and BIT(9)
fix ptp group bit mask to BIT(11) | BIT(12) | BIT(13)
add smi group with bit mask BIT(4)
[gregory: split the pcie group in 2, as at hardware level they can be
configured separately] Signed-off-by: Marek Behún <marek.behun@nic.cz> Signed-off-by: Gregory CLEMENT <gregory.clement@bootlin.com> Tested-by: Miquel Raynal <miquel.raynal@bootlin.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix multiple link training issues in aardvark driver. The main reason of
these issues was misunderstanding of what certain registers do, since their
names and comments were misleading: before commit 96be36dbffac ("PCI:
aardvark: Replace custom macros by standard linux/pci_regs.h macros"), the
pci-aardvark.c driver used custom macros for accessing standard PCIe Root
Bridge registers, and misleading comments did not help to understand what
the code was really doing.
After doing more tests and experiments I've come to the conclusion that the
SPEED_GEN register in aardvark sets the PCIe revision / generation
compliance and forces maximal link speed. Both GEN3 and GEN2 values set the
read-only PCI_EXP_FLAGS_VERS bits (PCIe capabilities version of Root
Bridge) to value 2, while GEN1 value sets PCI_EXP_FLAGS_VERS to 1, which
matches with PCI Express specifications revisions 3, 2 and 1 respectively.
Changing SPEED_GEN also sets the read-only bits PCI_EXP_LNKCAP_SLS and
PCI_EXP_LNKCAP2_SLS to corresponding speed.
(Note that PCI Express rev 1 specification does not define PCI_EXP_LNKCAP2
and PCI_EXP_LNKCTL2 registers and when SPEED_GEN is set to GEN1 (which
also sets PCI_EXP_FLAGS_VERS set to 1), lspci cannot access
PCI_EXP_LNKCAP2 and PCI_EXP_LNKCTL2 registers.)
Changing PCIe link speed can be done via PCI_EXP_LNKCTL2_TLS bits of
PCI_EXP_LNKCTL2 register. Armada 3700 Functional Specifications says that
the default value of PCI_EXP_LNKCTL2_TLS is based on SPEED_GEN value, but
tests showed that the default value is always 8.0 GT/s, independently of
speed set by SPEED_GEN. So after setting SPEED_GEN, we must also set value
in PCI_EXP_LNKCTL2 register via PCI_EXP_LNKCTL2_TLS bits.
Triggering PCI_EXP_LNKCTL_RL bit immediately after setting LINK_TRAINING_EN
bit actually doesn't do anything. Tests have shown that a delay is needed
after enabling LINK_TRAINING_EN bit. As triggering PCI_EXP_LNKCTL_RL
currently does nothing, remove it.
Commit 43fc679ced18 ("PCI: aardvark: Improve link training") introduced
code which sets SPEED_GEN register based on negotiated link speed from
PCI_EXP_LNKSTA_CLS bits of PCI_EXP_LNKSTA register. This code was added to
fix detection of Compex WLE900VX (Atheros QCA9880) WiFi GEN1 PCIe cards, as
otherwise these cards were "invisible" on PCIe bus (probably because they
crashed). But apparently more people reported the same issues with these
cards also with other PCIe controllers [1] and I was able to reproduce this
issue also with other "noname" WiFi cards based on Atheros QCA9890 chip
(with the same PCI vendor/device ids as Atheros QCA9880). So this is not an
issue in aardvark but rather an issue in Atheros QCA98xx chips. Also, this
issue only exists if the kernel is compiled with PCIe ASPM support, and a
generic workaround for this is to change PCIe Bridge to 2.5 GT/s link speed
via PCI_EXP_LNKCTL2_TLS_2_5GT bits in PCI_EXP_LNKCTL2 register [2], before
triggering PCI_EXP_LNKCTL_RL bit. This workaround also works when SPEED_GEN
is set to value GEN2 (5 GT/s). So remove this hack completely in the
aardvark driver and always set SPEED_GEN to value from 'max-link-speed' DT
property. Fix for Atheros QCA98xx chips is handled separately by patch [2].
These two things (code for triggering PCI_EXP_LNKCTL_RL bit and changing
SPEED_GEN value) also explain why commit 6964494582f5 ("PCI: aardvark:
Train link immediately after enabling training") somehow fixed detection of
those problematic Compex cards with Atheros chips: if triggering link
retraining (via PCI_EXP_LNKCTL_RL bit) was done immediately after enabling
link training (via LINK_TRAINING_EN), it did nothing. If there was a
specific delay, aardvark HW already initialized PCIe link and therefore
triggering link retraining caused the above issue. Compex cards triggered
link down event and disappeared from the PCIe bus.
Commit f4c7d053d7f7 ("PCI: aardvark: Wait for endpoint to be ready before
training link") added 100ms sleep before calling 'Start link training'
command and explained that it is a requirement of PCI Express
specification. But the code after this 100ms sleep was not doing 'Start
link training', rather it triggered PCI_EXP_LNKCTL_RL bit via PCIe Root
Bridge to put link into Recovery state.
The required delay after fundamental reset is already done in function
advk_pcie_wait_for_link() which also checks whether PCIe link is up.
So after removing the code which triggers PCI_EXP_LNKCTL_RL bit on PCIe
Root Bridge, there is no need to wait 100ms again. Remove the extra
msleep() call and update comment about the delay required by the PCI
Express specification.
According to Marvell Armada 3700 Functional Specifications, Link training
should be enabled via aardvark register LINK_TRAINING_EN after selecting
PCIe generation and x1 lane. There is no need to disable it prior resetting
card via PERST# signal. This disabling code was introduced in commit 5169a9851daa ("PCI: aardvark: Issue PERST via GPIO") as a workaround for
some Atheros cards. It turns out that this also is Atheros specific issue
and affects any PCIe controller, not only aardvark. Moreover this Atheros
issue was triggered by juggling with PCI_EXP_LNKCTL_RL, LINK_TRAINING_EN
and SPEED_GEN bits interleaved with sleeps. Now, after removing triggering
PCI_EXP_LNKCTL_RL, there is no need to explicitly disable LINK_TRAINING_EN
bit. So remove this code too. The problematic Compex cards described in
previous git commits are correctly detected in advk_pcie_train_link()
function even after applying all these changes.
Note that with this patch, and also prior this patch, some NVMe disks which
support PCIe GEN3 with 8 GT/s speed are negotiated only at the lowest link
speed 2.5 GT/s, independently of SPEED_GEN value. After manually triggering
PCI_EXP_LNKCTL_RL bit (e.g. from userspace via setpci), these NVMe disks
change link speed to 5 GT/s when SPEED_GEN was configured to GEN2. This
issue first needs to be properly investigated. I will send a fix in the
future.
On the other hand, some other GEN2 PCIe cards with 5 GT/s speed are
autonomously by HW autonegotiated at full 5 GT/s speed without need of any
software interaction.
Armada 3700 Functional Specifications describes the following steps for
link training: set SPEED_GEN to GEN2, enable LINK_TRAINING_EN, poll until
link training is complete, trigger PCI_EXP_LNKCTL_RL, poll until signal
rate is 5 GT/s, poll until link training is complete, enable ASPM L0s.
The requirement for triggering PCI_EXP_LNKCTL_RL can be explained by the
need to achieve 5 GT/s speed (as changing link speed is done by throw to
recovery state entered by PCI_EXP_LNKCTL_RL) or maybe as a part of enabling
ASPM L0s (but in this case ASPM L0s should have been enabled prior
PCI_EXP_LNKCTL_RL).
It is unknown why the original pci-aardvark.c driver was triggering
PCI_EXP_LNKCTL_RL bit before waiting for the link to be up. This does not
align with neither PCIe base specifications nor with Armada 3700 Functional
Specification. (Note that in older versions of aardvark, this bit was
called incorrectly PCIE_CORE_LINK_TRAINING, so this may be the reason.)
It is also unknown why Armada 3700 Functional Specification says that it is
needed to trigger PCI_EXP_LNKCTL_RL for GEN2 mode, as according to PCIe
base specification 5 GT/s speed negotiation is supposed to be entirely
autonomous, even if initial speed is 2.5 GT/s.
Change PCIe Max Payload Size setting in PCIe Device Control register to 512
bytes to align with PCIe Link Initialization sequence as defined in Marvell
Armada 3700 Functional Specification. According to the specification,
maximal Max Payload Size supported by this device is 512 bytes.
Without this kernel prints suspicious line:
pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 256 (was 16384, max 512)
With this change it changes to:
pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 256 (was 512, max 512)
Link: https://lore.kernel.org/r/20211005180952.6812-3-kabel@kernel.org Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver") Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Reviewed-by: Marek Behún <kabel@kernel.org> Cc: stable@vger.kernel.org Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In commit 6df6ba974a55 ("PCI: aardvark: Remove PCIe outbound window
configuration") was removed aardvark PCIe outbound window configuration and
commit description said that was recommended solution by HW designers.
But that commit completely removed support for configuring PCIe IO
resources without removing PCIe IO 'ranges' from DTS files. After that
commit PCIe IO space started to be treated as PCIe MEM space and accessing
it just caused kernel crash.
Moreover implementation of PCIe outbound windows prior that commit was
incorrect. It completely ignored offset between CPU address and PCIe bus
address and expected that in DTS is CPU address always same as PCIe bus
address without doing any checks. Also it completely ignored size of every
PCIe resource specified in 'ranges' DTS property and expected that every
PCIe resource has size 128 MB (also for PCIe IO range). Again without any
check. Apparently none of PCIe resource has in DTS specified size of 128
MB. So it was completely broken and thanks to how aardvark mask works,
configuration was completely ignored.
This patch reverts back support for PCIe outbound window configuration but
implementation is a new without issues mentioned above. PCIe outbound
window is required when DTS specify in 'ranges' property non-zero offset
between CPU and PCIe address space. To address recommendation by HW
designers as specified in commit description of 6df6ba974a55, set default
outbound parameters as PCIe MEM access without translation and therefore
for this PCIe 'ranges' it is not needed to configure PCIe outbound window.
For PCIe IO space is needed to configure aardvark PCIe outbound window.
This patch fixes kernel crash when trying to access PCIe IO space.
According to PCI Express Base Specifications (rev 4.0, 6.6.1
"Conventional reset"), after fundamental reset a 100ms delay is needed
prior to enabling link training.
Update comment in code to reflect this requirement.
Link: https://lore.kernel.org/r/20201202184659.3795-1-pali@kernel.org Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Move code which belongs to link training (delays and resets) into
advk_pcie_train_link() function, so everything related to link training,
including timings is at one place.
After experiments it can be observed that link training in aardvark
hardware is very sensitive to timings and delays, so it is a good idea to
have this code at the same place as link training calls.
This patch does not change behavior of aardvark initialization.
Link: https://lore.kernel.org/r/20200907111038.5811-6-pali@kernel.org Tested-by: Marek Behún <marek.behun@nic.cz> Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Include linux/gpio/consumer.h instead of linux/gpio.h, as is said in the
latter file.
This was reported by kernel test bot when compiling for s390.
drivers/pci/controller/pci-aardvark.c:350:2: error: implicit declaration of function 'gpiod_set_value_cansleep' [-Werror,-Wimplicit-function-declaration]
drivers/pci/controller/pci-aardvark.c:1074:21: error: implicit declaration of function 'devm_gpiod_get_from_of_node' [-Werror,-Wimplicit-function-declaration]
drivers/pci/controller/pci-aardvark.c:1076:14: error: use of undeclared identifier 'GPIOD_OUT_LOW'
Link: https://lore.kernel.org/r/202006211118.LxtENQfl%25lkp@intel.com Link: https://lore.kernel.org/r/20200907111038.5811-2-pali@kernel.org Fixes: 5169a9851daa ("PCI: aardvark: Issue PERST via GPIO") Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Reviewed-by: Marek Behún <marek.behun@nic.cz> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When there is no PCIe card connected and advk_pcie_rd_conf() or
advk_pcie_wr_conf() is called for PCI bus which doesn't belong to emulated
root bridge, the aardvark driver throws the following error message:
advk-pcie d0070000.pcie: config read/write timed out
Obviously accessing PCIe registers of disconnected card is not possible.
Extend check in advk_pcie_valid_device() function for validating
availability of PCIe bus. If PCIe link is down, then the device is marked
as Not Found and the driver does not try to access these registers.
This is just an optimization to prevent accessing PCIe registers when card
is disconnected. Trying to access PCIe registers of disconnected card does
not cause any crash, kernel just needs to wait for a timeout. So if card
disappear immediately after checking for PCIe link (before accessing PCIe
registers), it does not cause any problems.
Link: https://lore.kernel.org/r/20200702083036.12230-1-pali@kernel.org Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Most callers of config read do not check for return value. But most of the
ones that do, checks for error indication in 'val' variable.
This patch updates error handling in advk_pcie_rd_conf() function. If PIO
transfer fails then 'val' variable is set to 0xffffffff which indicates
failture.
Add support for issuing PERST via GPIO specified in 'reset-gpios'
property (as described in PCI device tree bindings).
Some buggy cards (e.g. Compex WLE900VX or WLE1216) are not detected
after reboot when PERST is not issued during driver initialization.
If bootloader already enabled link training then issuing PERST has no
effect for some buggy cards (e.g. Compex WLE900VX) and these cards are
not detected. We therefore clear the LINK_TRAINING_EN register before.
It was observed that Compex WLE900VX card needs to be in PERST reset
for at least 10ms if bootloader enabled link training.
Tested on Turris MOX.
Link: https://lore.kernel.org/r/20200430080625.26070-6-pali@kernel.org Tested-by: Tomasz Maciej Nowak <tmn505@gmail.com> Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently the aardvark driver trains link in PCIe gen2 mode. This may
cause some buggy gen1 cards (such as Compex WLE900VX) to be unstable or
even not detected. Moreover when ASPM code tries to retrain link second
time, these cards may stop responding and link goes down. If gen1 is
used this does not happen.
Unconditionally forcing gen1 is not a good solution since it may have
performance impact on gen2 cards.
To overcome this, read 'max-link-speed' property (as defined in PCI
device tree bindings) and use this as max gen mode. Then iteratively try
link training at this mode or lower until successful. After successful
link training choose final controller gen based on Negotiated Link Speed
from Link Status register, which should match card speed.
Link: https://lore.kernel.org/r/20200430080625.26070-5-pali@kernel.org Tested-by: Tomasz Maciej Nowak <tmn505@gmail.com> Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Marek Behún <marek.behun@nic.cz> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Reviewed-by: Rob Herring <robh@kernel.org> Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com> Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>