kernel/linux.git/drivers/char/hw_random, branch v6.18.21

hwrng: core - use RCU and work_struct to fix race condition

2026-02-26T22:59:06+00:00

[ Upstream commit cc2f39d6ac48e6e3cb2d6240bc0d6df839dd0828 ] Currently, hwrng_fill is not cleared until the hwrng_fillfn() thread exits. Since hwrng_unregister() reads hwrng_fill outside the rng_mutex lock, a concurrent hwrng_unregister() may call kthread_stop() again on the same task. Additionally, if hwrng_unregister() is called immediately after hwrng_register(), the stopped thread may have never been executed. Thus, hwrng_fill remains dirty even after hwrng_unregister() returns. In this case, subsequent calls to hwrng_register() will fail to start new threads, and hwrng_unregister() will call kthread_stop() on the same freed task. In both cases, a use-after-free occurs: refcount_t: addition on 0; use-after-free. WARNING: ... at lib/refcount.c:25 refcount_warn_saturate+0xec/0x1c0 Call Trace: kthread_stop+0x181/0x360 hwrng_unregister+0x288/0x380 virtrng_remove+0xe3/0x200 This patch fixes the race by protecting the global hwrng_fill pointer inside the rng_mutex lock, so that hwrng_fillfn() thread is stopped only once, and calls to kthread_run() and kthread_stop() are serialized with the lock held. To avoid deadlock in hwrng_fillfn() while being stopped with the lock held, we convert current_rng to RCU, so that get_current_rng() can read current_rng without holding the lock. To remove the lock from put_rng(), we also delay the actual cleanup into a work_struct. Since get_current_rng() no longer returns ERR_PTR values, the IS_ERR() checks are removed from its callers. With hwrng_fill protected by the rng_mutex lock, hwrng_fillfn() can no longer clear hwrng_fill itself. Therefore, if hwrng_fillfn() returns directly after current_rng is dropped, kthread_stop() would be called on a freed task_struct later. To fix this, hwrng_fillfn() calls schedule() now to keep the task alive until being stopped. The kthread_stop() call is also moved from hwrng_unregister() to drop_current_rng(), ensuring kthread_stop() is called on all possible paths where current_rng becomes NULL, so that the thread would not wait forever. Fixes: be4000bc4644 ("hwrng: create filler thread") Suggested-by: Herbert Xu Signed-off-by: Lianjie Wang Signed-off-by: Herbert Xu Signed-off-by: Sasha Levin

hwrng: core - Allow runtime disabling of the HW RNG

2026-02-26T22:59:06+00:00

[ Upstream commit e74b96d77da9eb5ee1b603c937c2adab5134a04b ] The HW RNG core allows for manual selection of which RNG device to use, but does not allow for no device to be enabled. It may be desirable to do this on systems with only a single suitable hardware RNG, where we need exclusive access to other functionality on this device. In particular when performing TPM firmware upgrades this lets us ensure the kernel does not try to access the device. Before: root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_* /sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0 /sys/devices/virtual/misc/hw_random/rng_current:tpm-rng-0 /sys/devices/virtual/misc/hw_random/rng_quality:1024 /sys/devices/virtual/misc/hw_random/rng_selected:0 After: root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_* /sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0 none /sys/devices/virtual/misc/hw_random/rng_current:tpm-rng-0 /sys/devices/virtual/misc/hw_random/rng_quality:1024 /sys/devices/virtual/misc/hw_random/rng_selected:0 root@debian-qemu-efi:~# echo none > /sys/devices/virtual/misc/hw_random/rng_current root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_* /sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0 none /sys/devices/virtual/misc/hw_random/rng_current:none grep: /sys/devices/virtual/misc/hw_random/rng_quality: No such device /sys/devices/virtual/misc/hw_random/rng_selected:1 (Observe using bpftrace no calls to TPM being made) root@debian-qemu-efi:~# echo "" > /sys/devices/virtual/misc/hw_random/rng_current root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_* /sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0 none /sys/devices/virtual/misc/hw_random/rng_current:tpm-rng-0 /sys/devices/virtual/misc/hw_random/rng_quality:1024 /sys/devices/virtual/misc/hw_random/rng_selected:0 (Observe using bpftrace that calls to the TPM resume) Signed-off-by: Jonathan McDowell Signed-off-by: Herbert Xu Stable-dep-of: cc2f39d6ac48 ("hwrng: core - use RCU and work_struct to fix race condition") Signed-off-by: Sasha Levin

hwrng: airoha - set rng quality to 900

2026-02-26T22:59:04+00:00

[ Upstream commit c0008a29a006091d7f9d288620c2456afa23ff27 ] Airoha uses RAW mode to collect noise from the TRNG. These appear to be unprocessed oscillations from the tero loop. For this reason, they do not have a perfect distribution and entropy. Simple noise compression reduces its size by 9%, so setting the quality to 900 seems reasonable. The same value is used by the downstream driver. Compare the size before and after compression: $ ls -l random_airoha* -rw-r--r-- 1 aleksander aleksander 76546048 Jan 3 23:43 random_airoha -rw-rw-r-- 1 aleksander aleksander 69783562 Jan 5 20:23 random_airoha.zip FIPS test results: $ cat random_airoha | rngtest -c 10000 rngtest 2.6 Copyright (c) 2004 by Henrique de Moraes Holschuh This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. rngtest: starting FIPS tests... rngtest: bits received from input: 200000032 rngtest: FIPS 140-2 successes: 0 rngtest: FIPS 140-2 failures: 10000 rngtest: FIPS 140-2(2001-10-10) Monobit: 9957 rngtest: FIPS 140-2(2001-10-10) Poker: 10000 rngtest: FIPS 140-2(2001-10-10) Runs: 10000 rngtest: FIPS 140-2(2001-10-10) Long run: 4249 rngtest: FIPS 140-2(2001-10-10) Continuous run: 0 rngtest: input channel speed: (min=953.674; avg=27698.935; max=19073.486)Mibits/s rngtest: FIPS tests speed: (min=59.791; avg=298.028; max=328.853)Mibits/s rngtest: Program run time: 647638 microseconds In general, these data look like real noise, but with lower entropy than expected. Fixes: e53ca8efcc5e ("hwrng: airoha - add support for Airoha EN7581 TRNG") Suggested-by: Benjamin Larsson Signed-off-by: Aleksander Jan Bajkowski Signed-off-by: Herbert Xu Signed-off-by: Sasha Levin

Merge tag 'v6.18-p1' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6

2025-10-04T21:59:29+00:00

Pull crypto updates from Herbert Xu: "Drivers: - Add ciphertext hiding support to ccp - Add hashjoin, gather and UDMA data move features to hisilicon - Add lz4 and lz77_only to hisilicon - Add xilinx hwrng driver - Add ti driver with ecb/cbc aes support - Add ring buffer idle and command queue telemetry for GEN6 in qat Others: - Use rcu_dereference_all to stop false alarms in rhashtable - Fix CPU number wraparound in padata" * tag 'v6.18-p1' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (78 commits) dt-bindings: rng: hisi-rng: convert to DT schema crypto: doc - Add explicit title heading to API docs hwrng: ks-sa - fix division by zero in ks_sa_rng_init KEYS: X.509: Fix Basic Constraints CA flag parsing crypto: anubis - simplify return statement in anubis_mod_init crypto: hisilicon/qm - set NULL to qm->debug.qm_diff_regs crypto: hisilicon/qm - clear all VF configurations in the hardware crypto: hisilicon - enable error reporting again crypto: hisilicon/qm - mask axi error before memory init crypto: hisilicon/qm - invalidate queues in use crypto: qat - Return pointer directly in adf_ctl_alloc_resources crypto: aspeed - Fix dma_unmap_sg() direction rhashtable: Use rcu_dereference_all and rcu_dereference_all_check crypto: comp - Use same definition of context alloc and free ops crypto: omap - convert from tasklet to BH workqueue crypto: qat - Replace kzalloc() + copy_from_user() with memdup_user() crypto: caam - double the entropy delay interval for retry padata: WQ_PERCPU added to alloc_workqueue users padata: replace use of system_unbound_wq with system_dfl_wq crypto: cryptd - WQ_PERCPU added to alloc_workqueue users ...

Merge tag 'sparc-for-6.18-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/alarsson/linux-sparc

2025-10-04T17:59:06+00:00

Pull sparc updates from Andreas Larsson: - Add relocation handling for R_SPARC_UA64 for sparc64 that is generated by llvm and clarify printout on missing relocation handler - Fix missing hugetlb tte initialization for sun4u - Code cleanup for redundant use of __GPF_NOWARN for sparc64 - Fix prototypes of reads[bwl]() for sparc64 by adding missing const and volatile pointer qualifiers - Fix bugs in accurate exception reporting in multiple machine specific sparc64 variants of copy_{from,to}_user() for sparc64 - Fix memory leak in error handling for sparc32 - Drop -ansi from asflags and replace __ASSEMBLY__ with __ASSEMBLER__ in headers for all arch/sparc - Replace strcpy() with strscpy() for all arch/sparc * tag 'sparc-for-6.18-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/alarsson/linux-sparc: (22 commits) sparc: Replace deprecated strcpy() with strscpy() in handle_nextprop_quirks() sparc64: Replace deprecated strcpy() with strscpy() in build_path_component() sparc: Replace deprecated strcpy() with strscpy() in prom_32.c sparc: Replace deprecated strcpy() with strscpy() in domain services driver sparc64: Replace deprecated strcpy() with strscpy() in prom_nextprop() sparc: floppy: Replace deprecated strcpy() with strscpy() in sun_floppy_init() sparc: parport: Replace deprecated strcpy() with strscpy() in ecpp_probe() sparc: PCI: Replace deprecated strcpy() with strscpy() sparc: Replace __ASSEMBLY__ with __ASSEMBLER__ in non-uapi headers sparc: Replace __ASSEMBLY__ with __ASSEMBLER__ in uapi headers sparc: Drop the "-ansi" from the asflags sparc: fix error handling in scan_one_device() sparc: fix accurate exception reporting in copy_{from,to}_user for M7 sparc: fix accurate exception reporting in copy_to_user for Niagara 4 sparc: fix accurate exception reporting in copy_{from_to}_user for Niagara sparc: fix accurate exception reporting in copy_{from_to}_user for UltraSPARC III sparc: fix accurate exception reporting in copy_{from_to}_user for UltraSPARC sparc64: fix prototypes of reads[bwl]() sparc64: Remove redundant __GFP_NOWARN sparc64: fix hugetlb for sun4u ...

hwrng: ks-sa - fix division by zero in ks_sa_rng_init

2025-09-28T03:54:48+00:00

Fix division by zero in ks_sa_rng_init caused by missing clock pointer initialization. The clk_get_rate() call is performed on an uninitialized clk pointer, resulting in division by zero when calculating delay values. Add clock initialization code before using the clock. Fixes: 6d01d8511dce ("hwrng: ks-sa - Add minimum sleep time before ready-polling") Signed-off-by: Nishanth Menon drivers/char/hw_random/ks-sa-rng.c | 7 +++++++ 1 file changed, 7 insertions(+) Reviewed-by: Alexander Sverdlin Signed-off-by: Herbert Xu

sparc: Replace ASSEMBLY with ASSEMBLER in non-uapi headers

2025-09-26T15:06:09+00:00

While the GCC and Clang compilers already define __ASSEMBLER__ automatically when compiling assembly code, __ASSEMBLY__ is a macro that only gets defined by the Makefiles in the kernel. This can be very confusing when switching between userspace and kernelspace coding, or when dealing with uapi headers that rather should use __ASSEMBLER__ instead. So let's standardize on the __ASSEMBLER__ macro that is provided by the compilers now. This is a completely mechanical patch (done with a simple "sed -i" statement). Cc: David S. Miller Cc: Andreas Larsson Cc: sparclinux@vger.kernel.org Signed-off-by: Thomas Huth Reviewed-by: Andreas Larsson Signed-off-by: Andreas Larsson

char: hw_random: atmel: make it selectable for ARCH_MICROCHIP

2025-09-09T08:34:22+00:00

LAN969x uses the Atmel HWRNG, so make it selectable for ARCH_MICROCHIP to avoid needing to update depends in future if other Microchip SoC-s use it as well. Signed-off-by: Robert Marko Acked-by: Daniel Machon Signed-off-by: Nicolas Ferre

hwrng: cn10k - Remove the use of dev_err_probe()

2025-08-30T07:43:26+00:00

Logging messages that show some type of "out of memory" error are generally unnecessary as there is a generic message and a stack dump done by the memory subsystem. These messages generally increase kernel size without much added value[1]. The dev_err_probe() doesn't do anything when error is '-ENOMEM'. Therefore, remove the useless call to dev_err_probe(), and just return the value instead. [1]: https://lore.kernel.org/lkml/1402419340.30479.18.camel@joe-AO725/ Signed-off-by: Liao Yuanhong Signed-off-by: Herbert Xu

hwrng: timeriomem - Use us_to_ktime() where appropriate

2025-08-16T09:24:31+00:00

It is better to replace ns_to_ktime() with us_to_ktime(), which can make the code clearer. Signed-off-by: Xichao Zhao Signed-off-by: Herbert Xu