Merge tag 'v6.16-p1' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6

Pull crypto updates from Herbert Xu:
 "API:
   - Fix memcpy_sglist to handle partially overlapping SG lists
   - Use memcpy_sglist to replace null skcipher
   - Rename CRYPTO_TESTS to CRYPTO_BENCHMARK
   - Flip CRYPTO_MANAGER_DISABLE_TEST into CRYPTO_SELFTESTS
   - Hide CRYPTO_MANAGER
   - Add delayed freeing of driver crypto_alg structures

  Compression:
   - Allocate large buffers on first use instead of initialisation in scomp
   - Drop destination linearisation buffer in scomp
   - Move scomp stream allocation into acomp
   - Add acomp scatter-gather walker
   - Remove request chaining
   - Add optional async request allocation

  Hashing:
   - Remove request chaining
   - Add optional async request allocation
   - Move partial block handling into API
   - Add ahash support to hmac
   - Fix shash documentation to disallow usage in hard IRQs

  Algorithms:
   - Remove unnecessary SIMD fallback code on x86 and arm/arm64
   - Drop avx10_256 xts(aes)/ctr(aes) on x86
   - Improve avx-512 optimisations for xts(aes)
   - Move chacha arch implementations into lib/crypto
   - Move poly1305 into lib/crypto and drop unused Crypto API algorithm
   - Disable powerpc/poly1305 as it has no SIMD fallback
   - Move sha256 arch implementations into lib/crypto
   - Convert deflate to acomp
   - Set block size correctly in cbcmac

  Drivers:
   - Do not use sg_dma_len before mapping in sun8i-ss
   - Fix warm-reboot failure by making shutdown do more work in qat
   - Add locking in zynqmp-sha
   - Remove cavium/zip
   - Add support for PCI device 0x17D8 to ccp
   - Add qat_6xxx support in qat
   - Add support for RK3576 in rockchip-rng
   - Add support for i.MX8QM in caam

  Others:
   - Fix irq_fpu_usable/kernel_fpu_begin inconsistency during CPU bring-up
   - Add new SEV/SNP platform shutdown API in ccp"

* tag 'v6.16-p1' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (382 commits)
  x86/fpu: Fix irq_fpu_usable() to return false during CPU onlining
  crypto: qat - add missing header inclusion
  crypto: api - Redo lookup on EEXIST
  Revert "crypto: testmgr - Add hash export format testing"
  crypto: marvell/cesa - Do not chain submitted requests
  crypto: powerpc/poly1305 - add depends on BROKEN for now
  Revert "crypto: powerpc/poly1305 - Add SIMD fallback"
  crypto: ccp - Add missing tee info reg for teev2
  crypto: ccp - Add missing bootloader info reg for pspv5
  crypto: sun8i-ce - move fallback ahash_request to the end of the struct
  crypto: octeontx2 - Use dynamic allocated memory region for lmtst
  crypto: octeontx2 - Initialize cptlfs device info once
  crypto: xts - Only add ecb if it is not already there
  crypto: lrw - Only add ecb if it is not already there
  crypto: testmgr - Add hash export format testing
  crypto: testmgr - Use ahash for generic tfm
  crypto: hmac - Add ahash support
  crypto: testmgr - Ignore EEXIST on shash allocation
  crypto: algapi - Add driver template support to crypto_inst_setname
  crypto: shash - Set reqsize in shash_alg
  ...

1 files changed, 21 insertions, 13 deletions

diff --git a/arch/x86/kernel/fpu/core.c b/arch/x86/kernel/fpu/core.c
index 91d6341f281f..399f43aa78d5 100644
--- a/arch/x86/kernel/fpu/core.c
+++ b/arch/x86/kernel/fpu/core.c
@@ -43,8 +43,11 @@ struct fpu_state_config fpu_user_cfg __ro_after_init;
  */
 struct fpstate init_fpstate __ro_after_init;
 
-/* Track in-kernel FPU usage */
-static DEFINE_PER_CPU(bool, in_kernel_fpu);
+/*
+ * Track FPU initialization and kernel-mode usage. 'true' means the FPU is
+ * initialized and is not currently being used by the kernel:
+ */
+DEFINE_PER_CPU(bool, kernel_fpu_allowed);
 
 /*
  * Track which context is using the FPU on the CPU:
@@ -61,15 +64,18 @@ bool irq_fpu_usable(void)
 		return false;
 
 	/*
-	 * In kernel FPU usage already active?  This detects any explicitly
-	 * nested usage in task or softirq context, which is unsupported.  It
-	 * also detects attempted usage in a hardirq that has interrupted a
-	 * kernel-mode FPU section.
+	 * Return false in the following cases:
+	 *
+	 * - FPU is not yet initialized. This can happen only when the call is
+	 *   coming from CPU onlining, for example for microcode checksumming.
+	 * - The kernel is already using the FPU, either because of explicit
+	 *   nesting (which should never be done), or because of implicit
+	 *   nesting when a hardirq interrupted a kernel-mode FPU section.
+	 *
+	 * The single boolean check below handles both cases:
 	 */
-	if (this_cpu_read(in_kernel_fpu)) {
-		WARN_ON_FPU(!in_hardirq());
+	if (!this_cpu_read(kernel_fpu_allowed))
 		return false;
-	}
 
 	/*
 	 * When not in NMI or hard interrupt context, FPU can be used in:
@@ -431,9 +437,10 @@ void kernel_fpu_begin_mask(unsigned int kfpu_mask)
 		fpregs_lock();
 
 	WARN_ON_FPU(!irq_fpu_usable());
-	WARN_ON_FPU(this_cpu_read(in_kernel_fpu));
 
-	this_cpu_write(in_kernel_fpu, true);
+	/* Toggle kernel_fpu_allowed to false: */
+	WARN_ON_FPU(!this_cpu_read(kernel_fpu_allowed));
+	this_cpu_write(kernel_fpu_allowed, false);
 
 	if (!(current->flags & (PF_KTHREAD | PF_USER_WORKER)) &&
 	    !test_thread_flag(TIF_NEED_FPU_LOAD)) {
@@ -453,9 +460,10 @@ EXPORT_SYMBOL_GPL(kernel_fpu_begin_mask);
 
 void kernel_fpu_end(void)
 {
-	WARN_ON_FPU(!this_cpu_read(in_kernel_fpu));
+	/* Toggle kernel_fpu_allowed back to true: */
+	WARN_ON_FPU(this_cpu_read(kernel_fpu_allowed));
+	this_cpu_write(kernel_fpu_allowed, true);
 
-	this_cpu_write(in_kernel_fpu, false);
 	if (!irqs_disabled())
 		fpregs_unlock();
 }