The idea that you can arbitarily read 32-bits from PCI configuration space,
modify a sub-field (like the command register) and write it back without
consequence is deeply flawed.
Status registers (such as the status register, PCIe device status register,
etc) contain status bits which are read, write-one-to-clear.
What this means is that reading 32-bits from the command register,
modifying the command register, and then writing it back has the effect of
clearing any status bits that were indicating at that time. Same for the
PCIe device control register clearing bits in the PCIe device status
register.
Since the Armada chips support byte, 16-bit and 32-bit accesses to the
registers (unless otherwise stated) and the PCI configuration data register
does not specify otherwise, it seems logical that the chip can indeed
generate the proper configuration access cycles down to byte level.
Testing with an ASM1062 PCIe to SATA mini-PCIe card on Armada 388. PCIe
capability at 0x80, DevCtl at 0x88, DevSta at 0x8a.
Before:
/# setpci -s 1:0.0 0x88.l - DevSta: CorrErr+
00012810
/# setpci -s 1:0.0 0x88.w=0x2810 - Write DevCtl only
/# setpci -s 1:0.0 0x88.l - CorrErr cleared - FAIL
00002810
After:
/# setpci -s 1:0.0 0x88.l - DevSta: CorrErr+
00012810
/# setpci -s 1:0.0 0x88.w=0x2810 - check DevCtl only write
/# setpci -s 1:0.0 0x88.l - CorErr remains set
00012810
/# setpci -s 1:0.0 0x88.w=0x281f - check DevCtl write works
/# setpci -s 1:0.0 0x88.l - devctl field updated
0001281f
/# setpci -s 1:0.0 0x8a.w=0xffff - clear DevSta
/# setpci -s 1:0.0 0x88.l - CorrErr now cleared
0000281f
/# setpci -s 1:0.0 0x88.w=0x2810 - restore DevCtl
/# setpci -s 1:0.0 0x88.l - check
00002810
[bhelgaas: changelog]
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> (Armada XP GP)
Tested-by: Andrew Lunn <andrew@lunn.ch> (Kirkwood DIR665)
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
PCI requires reads to reserved or unimplemented configuration space to
return zero and complete normally (see PCI r3.0, sec 6.1). However, the
root port software implementation was returning 0xfffffff and
PCIBIOS_BAD_REGISTER_NUMBER.
Return zero when reading reserved or unimplemented config space.
[bhelgaas: changelog]
Tested-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> (Armada XP GP)
Tested-by: Andrew Lunn <andrew@lunn.ch> (Kirkwood DIR665)
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
For some reason, only the little-endian flavor of
powerpc provided the zero_bytemask() implementation.
Reported-by: Michal Sojka <sojkam1@fel.cvut.cz>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>
- properly get the slow clock from timer-atmel-st, tcb_clksrc and pwm-atmel-tcb
- small fix in an error path for tcb_clksrc
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iQIcBAABCAAGBQJWE+cSAAoJEKbNnwlvZCyzxtYP/2UiUB2aHcHiFwINhiyOr6p/
HmhJvxR0A72yq8jUkC5UEozj/U4boA66v+u/fR5DB8xV8Fj9VCoflrCqmC0VNoqT
Vh8RklpO0OauiosLDC+pKlZxNHcyc/hDNCv+IeV4VIJKk6dOj+vpaHRqgYljk4/m
zsUXZPKku1DZ45sQwx5a2FqEpsXlMcyawjXPZFuPHymlB2PPKiQDtvouIHOUyDoO
dvlX7C7MVR5XaIiG8pSHMMmNNYhlrGnTYgUEEo0DhqKXpQLLo4XbCxdsq2hGIDhT
XoXgqQVKTLVha3/tHnJjy43OqO65ptTKKMvyFyXaNhknLAz8aomJAgZXeOJSWYUB
a6Z2NRKf28do7+57C95hkbWMNiTziNud903DIuK3S/6c5EmpjJ1IJupGUp4JNmkW
FfT+r4s1J3sO9QNG9Y86/P2jTq3f0aIJesMpQ5mU4xkzVfFZCKeXUOylNLR37VE+
8ofgVyC8kbqClGEuePR0KvuinIj7xGlGOmSNR+M5FR4sZVEvc6xHVwnDXcT9Hbpe
pvVGbXPNfhEEP0E2gqf1GFcqAx8bYQzeFOh0CSA8sEYqP+c8GgfFIgjeP31W4Gti
gjgYeAAUg8i45WZBCPYAJbpsDwzxv8vvpDvioqVJDdQo+aJv0TA6ITl9eHHj0mgV
8WpYxQWiZ2PcBNGtqteK
=UbC0
-----END PGP SIGNATURE-----
Merge tag 'at91-cleanup-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux into next/drivers
Merge "First batch of cleanups for 4.4:" from Alexandre Belloni:
- properly get the slow clock from timer-atmel-st, tcb_clksrc and pwm-atmel-tcb
- small fix in an error path for tcb_clksrc
* tag 'at91-cleanup-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux:
misc: atmel_tclib: get and use slow clock
clocksource: tcb_clksrc: fix setup_clkevents error path
clocksource: atmel-st: get and use slow clock
If the bus is being configured with a bus-range that does not start at
zero, pass that starting bus number to pci_scan_root_bus(). Passing the
incorrect value of zero causes attempted config accesses outside of the
supported range, which cascades to an OOPs spew and eventual kernel panic.
Signed-off-by: David Daney <david.daney@cavium.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Will Deacon <will.deacon@arm.com>
DaVinci clock code. And a fix to use a more appropiate
format specifier in a debug message.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)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=Pbm7
-----END PGP SIGNATURE-----
Merge tag 'davinci-for-v4.4/fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/nsekhar/linux-davinci into next/fixes-non-critical
Merge "DaVinci non-critical fixes for v4.4" from Sekhar Nori:
Fix for incorrect handling of NULL clk pointer in
DaVinci clock code. And a fix to use a more appropiate
format specifier in a debug message.
* tag 'davinci-for-v4.4/fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/nsekhar/linux-davinci:
ARM: davinci: clock: Correct return values for API functions
ARM: davinci: re-use %*ph specifier
Tronfy is an emerging brand in China specializing in Home Theater
solutions for the normal consumers.
Signed-off-by: Carlo Caione <carlo@endlessm.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
* Add missing CPG/MSTP Clock Domain for sound on r8a779[01] SoCs
* Tidy up SCI resource region on r8a779[018] SoCs
* Add pinmux for iic0 on Lager board
* Use CCF for audio clock on Lager and Koelsch boards
* Use serial0 and 1 as serial ports on Marzen board
* Use adxl345-specific compatible property for KZM9G board
* Document compat string for Silk board
* Enable GPIO, I2C, PCI, QSPI, USB PHY and HS, and VIN support on r8a7794/Silk
* Add initial support for r8a7791/porter
* Add common file for AA121TD01 panel
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iQIcBAABAgAGBQJWFHsPAAoJENfPZGlqN0++WBQQAKewTnv8GBqBtgT7WFz3ADLV
CSDqME9lx70Hwfst8PWS0MLD4Fs/g/thf36LUbYSomZ35hzy4FGCXHvEOSwjZ9wq
zEvwoIu9yWv3Ujc2EZ07YndXhqliyaNwRkC9foRHx3Nzasaa+2m4mDXIF6ENndlt
9hJoWuE+iAAsTFAkcTp+0Bu97VNrtlcEjEA59KM5xGEBdCfN3mVt5+mHGcQh4yHF
rBFkcqC7+8y5Igxk2VrcfMIlPv1cxWy/llCRQrvyV48CLf1HyIHYtfKx0Bj2JMFL
pH841A6Sxi9Uhce0fjr4YCkBFpsGRtXjWOovyMKM+042PS7ugOUqf7GMskqIoLkN
6hLzsNR2Da8p3OQrhXfv1dHbLkUMkyRkE6n2btAPnJWxAQwQBtTgLUXq561NnJuD
to52gIKnmH9DeW3uaslZSE/GlgAeusQI9QwUIVnGq6m1AT9t3UKNaoohRo68pjQC
P6YACrZLys0luPQkwGoZJdBH9clI4OBkcmw9t/6N7yg10aNnME9yjWaEbk/KsGVy
yyiiqm2AKhXM6Qsho4DA7tSoxQRzX8Mee43vRRLlOIothzuVVdSGx/61uJtWLWFc
52wNBdmSBWkaUTw237K7fXkIUtKXlhruCW1At3gBbBY2S53jOBrJ5+ZBFDf2hRpm
GecycolgMWa+8/SID/jc
=ceDx
-----END PGP SIGNATURE-----
Merge tag 'renesas-dt-for-v4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas into next/dt
Merge "Renesas ARM Based SoC DT Updates for v4.4" from Simon Horman:
* Add missing CPG/MSTP Clock Domain for sound on r8a779[01] SoCs
* Tidy up SCI resource region on r8a779[018] SoCs
* Add pinmux for iic0 on Lager board
* Use CCF for audio clock on Lager and Koelsch boards
* Use serial0 and 1 as serial ports on Marzen board
* Use adxl345-specific compatible property for KZM9G board
* Document compat string for Silk board
* Enable GPIO, I2C, PCI, QSPI, USB PHY and HS, and VIN support on r8a7794/Silk
* Add initial support for r8a7791/porter
* Add common file for AA121TD01 panel
* tag 'renesas-dt-for-v4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas: (28 commits)
ARM: shmobile: porter: add Ether DT support
ARM: shmobile: fix SILK board name
ARM: shmobile: r8a7794: add HS-USB DT support
ARM: shmobile: dts: Add common file for AA121TD01 panel
ARM: shmobile: r8a7794: link PCI USB devices to USB PHY
ARM: shmobile: silk: enable USB PHY
ARM: shmobile: r8a7794: add USB PHY DT support
ARM: shmobile: porter: initial device tree
ARM: shmobile: add Porter board DT bindings
ARM: shmobile: silk: enable internal PCI
ARM: shmobile: r8a7794: add internal PCI bridge nodes
ARM: shmobile: r8a7790: lager: add pinmux for iic0
ARM: shmobile: r8a7778: tidyup SSI resource region
ARM: shmobile: r8a7791: tidyup SSI resource region
ARM: shmobile: r8a7790: tidyup SSI resource region
ARM: shmobile: lager: use CCF for audio clock
ARM: shmobile: koelsch: use CCF for audio clock
ARM: shmobile: silk: add VIN0/ADV7180 DT support
ARM: shmobile: r8a7794: add VIN DT support
ARM: shmobile: silk: add I2C1 DT support
...
The generic driver kept a global struct pci_ops ("gen_pci_ops") which it
patched with the .map_bus() method appropriate for the bus device. This is
a problem when we have two different types of bus devices: the .map_bus()
method for the last device probed clobbers the method for previous devices.
The result is that only the last bus device probed has the correct
.map_bus(), and the others fail.
Move the struct pci_ops into the bus-specific structure and initialize a
pointer to it when the bus device is probed.
Signed-off-by: David Daney <david.daney@cavium.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Will Deacon <will.deacon@arm.com>
We were getting build warning about "Section mismatch".
dmi_platform_intel_broadwell is being referenced from the probe function
rt5645_i2c_probe(), but dmi_platform_intel_broadwell was marked with
__initdata.
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Reviewed-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Add a driver for the SPDIF transceiver available on RK3066, RK3188 and
RK3288. Heavily based on the rockchip i2s driver.
Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Mark Brown <broonie@kernel.org>
Add devicetree bindings for the spdif tranceiver found on
found on rk3066, rk3188 and rk3288 SoCs
Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>
Signed-off-by: Mark Brown <broonie@kernel.org>
* sunrpc:
SUNRPC: Use MSG_SENDPAGE_NOTLAST in xs_send_pagedata()
SUNRPC: Move AF_LOCAL receive data path into a workqueue context
SUNRPC: Move UDP receive data path into a workqueue context
SUNRPC: Move TCP receive data path into a workqueue context
SUNRPC: Refactor TCP receive
For symmetry with the synchronous handler, and so that we can potentially
handle errors such as NFS4ERR_BADNAME.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
The cns3xxx_pcie_hw_init function uses excessive kernel
stack space because of a hack that puts a fake struct
pci_sys_data and struct pci_bus on the stack in order to
call the generic pci_bus_read_config accessors, which causes
a warning in ARM allmodconfig builds:
arch/arm/mach-cns3xxx/pcie.c:266:1: warning: the frame size of 1080 bytes is larger than 1024 bytes
I've spent a few hours trying to find out what exactly this
code is wants to achieve here. The obvious part is setting
up the host_regs using config space accessors, and this can
simply be changed to use direct MMIO accesses, as I do
in this patch.
The second part is how the driver sets up the Max_Read_Request_Size
value for the first device/function on bus 1, i.e. the device
plugged directly into the PCIe root port.
For all I can tell, this is in fact incomplete, as it does not
perform the same setting on devices attached to a PCIe switch,
or multi-function devices.
The solution for this part fortunately is even easier: if we
just set the global pcie_bus_config variable to PCIE_BUS_PEER2PEER,
all PCIe devices in the system are limited to 128 byte MPS, which
in turn limits the MRRS to 128 bytes for all devices, and we
no longer even need to touch any devices.
With those two changes in place, we no longer need the fake
pci_sys_data/pci_bus structures for faking config space writes,
and the stack usage goes down as well.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Krzysztof Halasa <khalasa@piap.pl>
Use goto to handle the error path to avoid duplicating the same code. In
the error path intel_dig_port is the last one to be released as it was
the first one to be allocated and ideally the error path should be the
reverse of the execution path.
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The original representation of FIFO size in the driver coupled with the
ambiguity in the documentation meant that it was easy to confuse readers.
This lead to a false positive BUG-find and subsequently time wastage
debugging this phantom issue.
Hopefully this patch can prevent future readers from falling into the
same trap.
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Samples are documented to be available every 0.667us, so in theory
the 8 sample deep FIFO should take 5.336us to fill. However, during
thorough testing, it became apparent that filling the FIFO actually
takes closer to 12us.
Also take into consideration that udelay() can behave oddly i.e. not
delay for as long as requested.
Suggested-by: Russell King <rmk+kernel@arm.linux.org.uk>:
"IIRC, Linus recommends a x2 factor on delays, especially
timeouts generated by these functions.
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
DT nodes should not append their addresses with '0x'.
Suggested-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
clk_prepare_enable() can fail so add a check for this and
return the error code if it fails.
Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Add support for AES working in XEX-based Tweaked-codebook mode with
ciphertext Stealing (XTS)
sector index - HW limitation: CAAM device supports sector index of only
8 bytes to be used for sector index inside IV, instead of whole 16 bytes
received on request. This represents 2 ^ 64 = 16,777,216 Tera of possible
values for sector index.
Signed-off-by: Cristian Hristea <cristi.hristea@gmail.com>
Signed-off-by: Horia Geanta <horia.geanta@freescale.com>
Signed-off-by: Alex Porosanu <alexandru.porosanu@freescale.com>
Signed-off-by: Catalin Vasile <catalin.vasile@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The qce driver use two dma_map_sg path according to SG are chained
or not.
Since dma_map_sg can handle both case, clean the code with all
references to sg chained.
Thus removing qce_mapsg, qce_unmapsg and qce_countsg functions.
Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The convention is to use the name of the module in the driver structures
that are used for registering the device. The CCP module is currently
using a descriptive name. Replace the descriptive name with module name.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The CCP is meant to be more of an offload engine than an accelerator
engine. To avoid any confusion, change references to accelerator to
offload.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
With the creation of the device_dma_is_coherent API the "use_acpi" field
is no longer needed, so remove it.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Replace the usage of BUG_ON with WARN_ON and return an error.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
We need to explicitly check the AVX and AES CPU features, as we can't
infer them from the related XSAVE feature flags. For example, the
Core i3 2310M passes the XSAVE feature test but does not implement
AES-NI.
Reported-and-tested-by: Stéphane Glondu <glondu@debian.org>
References: https://bugs.debian.org/800934
Fixes: ce4f5f9b65 ("x86/fpu, crypto x86/camellia_aesni_avx: Simplify...")
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Cc: stable <stable@vger.kernel.org> # 4.2
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Some of the crypto algorithms write to the initialization vector,
but no space has been allocated for it. This clobbers adjacent memory.
Cc: stable@vger.kernel.org
Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Recursive dependency issues with kconfig are unavoidable due to
some limitations with kconfig, since these issues are recurring
provide a hint to the user how they can resolve these dependency
issues and also document why such limitation exists.
While at it also document a bit of future prospects of ways to
enhance Kconfig, including providing formal semantics and evaluation
of use of a SAT solver. If you're interested in this work or prospects
of it check out the kconfig-sat project wiki [0] and mailing list [1].
[0] http://kernelnewbies.org/KernelProjects/kconfig-sat
[1] https://groups.google.com/d/forum/kconfig-sat
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: James Bottomley <jbottomley@odin.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Bolle <pebolle@tiscali.nl>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Mate Soos <soos.mate@gmail.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Michal Marek <mmarek@suse.com>
If we're sending more than one page via kernel_sendpage(), then set
MSG_SENDPAGE_NOTLAST between the pages so that we don't send suboptimal
frames (see commit 2f53384424 and commit 35f9c09fe9).
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
When the controller is unconfigured (for example it does not have a
valid Bluetooth address), then the basic debugfs entries for dut_mode
and vendor_diag are not creates. Ensure they are created in __hci_init
and also __hci_unconf_init functions. One of them is called during setup
stage of a new controller.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Stream protocols such as TCP can often build up a backlog of data to be
read due to ordering. Combine this with the fact that some workloads such
as NFS read()-intensive workloads need to receive a lot of data per RPC
call, and it turns out that receiving the data from inside a softirq
context can cause starvation.
The following patch moves the TCP data receive into a workqueue context.
We still end up calling tcp_read_sock(), but we do so from a process
context, meaning that softirqs are enabled for most of the time.
With this patch, I see a doubling of read bandwidth when running a
multi-threaded iozone workload between a virtual client and server setup.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Move the TCP data receive loop out of xs_tcp_data_ready(). Doing so
will allow us to move the data receive out of the softirq context in
a set of followup patches.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Daniel Borkmann says:
====================
BPF/random32 updates
BPF update to split the prandom state apart, and to move the
*once helpers to the core. For details, please see individual
patches. Given the changes and since it's in the tree for
quite some time, net-next is a better choice in our opinion.
v1 -> v2:
- Make DO_ONCE() type-safe, remove the kvec helper. Credits
go to Alexei Starovoitov for the __VA_ARGS__ hint, thanks!
- Add a comment to the DO_ONCE() helper as suggested by Alexei.
- Rework prandom_init_once() helper to the new API.
- Keep Alexei's Acked-by on the last patch.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
While recently arguing on a seccomp discussion that raw prandom_u32()
access shouldn't be exposed to unpriviledged user space, I forgot the
fact that SKF_AD_RANDOM extension actually already does it for some time
in cBPF via commit 4cd3675ebf ("filter: added BPF random opcode").
Since prandom_u32() is being used in a lot of critical networking code,
lets be more conservative and split their states. Furthermore, consolidate
eBPF and cBPF prandom handlers to use the new internal PRNG. For eBPF,
bpf_get_prandom_u32() was only accessible for priviledged users, but
should that change one day, we also don't want to leak raw sequences
through things like eBPF maps.
One thought was also to have own per bpf_prog states, but due to ABI
reasons this is not easily possible, i.e. the program code currently
cannot access bpf_prog itself, and copying the rnd_state to/from the
stack scratch space whenever a program uses the prng seems not really
worth the trouble and seems too hacky. If needed, taus113 could in such
cases be implemented within eBPF using a map entry to keep the state
space, or get_random_bytes() could become a second helper in cases where
performance would not be critical.
Both sides can trigger a one-time late init via prandom_init_once() on
the shared state. Performance-wise, there should even be a tiny gain
as bpf_user_rnd_u32() saves one function call. The PRNG needs to live
inside the BPF core since kernels could have a NET-less config as well.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
Cc: Chema Gonzalez <chema@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a prandom_init_once() facility that works on the rnd_state, so that
users that are keeping their own state independent from prandom_u32() can
initialize their taus113 per cpu states.
The motivation here is similar to net_get_random_once(): initialize the
state as late as possible in the hope that enough entropy has been
collected for the seeding. prandom_init_once() makes use of the recently
introduced prandom_seed_full_state() helper and is generic enough so that
it could also be used on fast-paths due to the DO_ONCE().
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
