- 27 8月, 2015 1 次提交
-
-
由 Ard Biesheuvel 提交于
When a task calls execve(), its FP/SIMD state is flushed so that none of the original program state is observeable by the incoming program. However, since this flushing consists of setting the in-memory copy of the FP/SIMD state to all zeroes, the CPU field is set to CPU 0 as well, which indicates to the lazy FP/SIMD preserve/restore code that the FP/SIMD state does not need to be reread from memory if the task is scheduled again on CPU 0 without any other tasks having entered userland (or used the FP/SIMD in kernel mode) on the same CPU in the mean time. If this happens, the FP/SIMD state of the old program will still be present in the registers when the new program starts. So set the CPU field to the invalid value of NR_CPUS when performing the flush, by calling fpsimd_flush_task_state(). Cc: <stable@vger.kernel.org> Reported-by: NChunyan Zhang <chunyan.zhang@spreadtrum.com> Reported-by: NJanet Liu <janet.liu@spreadtrum.com> Signed-off-by: NArd Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: NWill Deacon <will.deacon@arm.com>
-
- 12 6月, 2015 1 次提交
-
-
由 Janet Liu 提交于
Now FPSIMD don't handle HOTPLUG_CPU. This introduces bug after cpu down/up process. After cpu down/up process, the FPSMID hardware register is default value, not any process's fpsimd context. when CPU_DEAD set cpu's fpsimd_state to NULL, it will force to load the fpsimd context for the thread, to avoid the chance to skip to load the context. If process A is the last user process on CPU N before cpu down, and the first user process on the same CPU N after cpu up, A's fpsimd_state.cpu is the current cpu id, and per_cpu(fpsimd_last_state) points A's fpsimd_state, so kernel will not reload the context during it return to user space. Signed-off-by: NJanet Liu <janet.liu@spreadtrum.com> Signed-off-by: NXiongshan An <xiongshan.an@spreadtrum.com> Signed-off-by: NChunyan Zhang <chunyan.zhang@spreadtrum.com> [catalin.marinas@arm.com: some mostly cosmetic clean-ups] Signed-off-by: NCatalin Marinas <catalin.marinas@arm.com>
-
- 01 9月, 2014 1 次提交
-
-
由 Leo Yan 提交于
Now arm64 defers reloading FPSIMD state, but this optimization also introduces the bug after cpu resume back from low power mode. The reason is after the cpu has been powered off, s/w need set the cpu's fpsimd_last_state to NULL so that it will force to reload FPSIMD state for the thread, otherwise there has the chance to meet the condition for both the task's fpsimd_state.cpu field contains the id of the current cpu, and the cpu's fpsimd_last_state per-cpu variable points to the task's fpsimd_state, so finally kernel will skip to reload the context during it return back to userland. Acked-by: NArd Biesheuvel <ard.biesheuvel@linaro.org> Reviewed-by: NCatalin Marinas <catalin.marinas@arm.com> Signed-off-by: NLeo Yan <leoy@marvell.com> Signed-off-by: NWill Deacon <will.deacon@arm.com>
-
- 08 5月, 2014 3 次提交
-
-
由 Ard Biesheuvel 提交于
This patch modifies kernel_neon_begin() and kernel_neon_end(), so they may be called from any context. To address the case where only a couple of registers are needed, kernel_neon_begin_partial(u32) is introduced which takes as a parameter the number of bottom 'n' NEON q-registers required. To mark the end of such a partial section, the regular kernel_neon_end() should be used. Signed-off-by: NArd Biesheuvel <ard.biesheuvel@linaro.org>
-
由 Ard Biesheuvel 提交于
If a task gets scheduled out and back in again and nothing has touched its FPSIMD state in the mean time, there is really no reason to reload it from memory. Similarly, repeated calls to kernel_neon_begin() and kernel_neon_end() will preserve and restore the FPSIMD state every time. This patch defers the FPSIMD state restore to the last possible moment, i.e., right before the task returns to userland. If a task does not return to userland at all (for any reason), the existing FPSIMD state is preserved and may be reused by the owning task if it gets scheduled in again on the same CPU. This patch adds two more functions to abstract away from straight FPSIMD register file saves and restores: - fpsimd_restore_current_state -> ensure current's FPSIMD state is loaded - fpsimd_flush_task_state -> invalidate live copies of a task's FPSIMD state Signed-off-by: NArd Biesheuvel <ard.biesheuvel@linaro.org>
-
由 Ard Biesheuvel 提交于
There are two tacit assumptions in the FPSIMD handling code that will no longer hold after the next patch that optimizes away some FPSIMD state restores: . the FPSIMD registers of this CPU contain the userland FPSIMD state of task 'current'; . when switching to a task, its FPSIMD state will always be restored from memory. This patch adds the following functions to abstract away from straight FPSIMD register file saves and restores: - fpsimd_preserve_current_state -> ensure current's FPSIMD state is saved - fpsimd_update_current_state -> replace current's FPSIMD state Where necessary, the signal handling and fork code are updated to use the above wrappers instead of poking into the FPSIMD registers directly. Signed-off-by: NArd Biesheuvel <ard.biesheuvel@linaro.org>
-
- 17 12月, 2013 1 次提交
-
-
由 Lorenzo Pieralisi 提交于
When a CPU enters a low power state, its FP register content is lost. This patch adds a notifier to save the FP context on CPU shutdown and restore it on CPU resume. The context is saved and restored only if the suspending thread is not a kernel thread, mirroring the current context switch behaviour. Signed-off-by: NLorenzo Pieralisi <lorenzo.pieralisi@arm.com>
-
- 28 9月, 2013 1 次提交
-
-
由 Jiang Liu 提交于
If context switching happens during executing fpsimd_flush_thread(), stale value in FPSIMD registers will be saved into current thread's fpsimd_state by fpsimd_thread_switch(). That may cause invalid initialization state for the new process, so disable preemption when executing fpsimd_flush_thread(). Signed-off-by: NJiang Liu <jiang.liu@huawei.com> Cc: Jiang Liu <liuj97@gmail.com> Signed-off-by: NCatalin Marinas <catalin.marinas@arm.com>
-
- 20 8月, 2013 1 次提交
-
-
由 Ard Biesheuvel 提交于
Add <asm/neon.h> containing kernel_neon_begin/kernel_neon_end function declarations and corresponding definitions in fpsimd.c These are needed to wrap uses of NEON in kernel mode. The names are identical to the ones used in arm/ so code using intrinsics or vectorized by GCC can be shared between arm and arm64. Signed-off-by: NArd Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: NCatalin Marinas <catalin.marinas@arm.com>
-
- 17 9月, 2012 1 次提交
-
-
由 Catalin Marinas 提交于
This patch adds support for FP/ASIMD register bank saving and restoring during context switch and FP exception handling to generate SIGFPE. There are 32 128-bit registers and the context switching is currently done non-lazily. Benchmarks on real hardware are required before implementing lazy FP state saving/restoring. Signed-off-by: NWill Deacon <will.deacon@arm.com> Signed-off-by: NCatalin Marinas <catalin.marinas@arm.com> Acked-by: NTony Lindgren <tony@atomide.com> Acked-by: NArnd Bergmann <arnd@arndb.de> Acked-by: NNicolas Pitre <nico@linaro.org> Acked-by: NOlof Johansson <olof@lixom.net> Acked-by: NSantosh Shilimkar <santosh.shilimkar@ti.com>
-