From: Thomas Gleixner <tglx@linutronix.de>
To: LKML <linux-kernel@vger.kernel.org>
Cc: Andy Lutomirski <luto@kernel.org>,
Dave Hansen <dave.hansen@linux.intel.com>,
Fenghua Yu <fenghua.yu@intel.com>,
Tony Luck <tony.luck@intel.com>,
Yu-cheng Yu <yu-cheng.yu@intel.com>,
Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
Borislav Petkov <bp@suse.de>,
Peter Zijlstra <peterz@infradead.org>,
Kan Liang <kan.liang@linux.intel.com>
Subject: [patch 11/41] x86/fpu: Get rid of copy_supervisor_to_kernel()
Date: Fri, 11 Jun 2021 18:15:34 +0200 [thread overview]
Message-ID: <20210611163112.153392869@linutronix.de> (raw)
In-Reply-To: 20210611161523.508908024@linutronix.de
If the fast path of restoring the FPU state on sigreturn fails or is not
taken and the current task's FPU is active then the FPU has to be
deactivated for the slow path to allow a safe update of the tasks FPU
memory state.
With supervisor states enabled, this requires to save the supervisor state
in the memory state first. Supervisor states require XSAVES so saving only
the supervisor state requires to reshuffle the memory buffer because XSAVES
uses the compacted format and therefore stores the supervisor states at the
beginning of the memory state. That's just an overengineered optimization.
Get rid of it and save the full state for this case.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
V4: New patch
---
arch/x86/include/asm/fpu/xstate.h | 1
arch/x86/kernel/fpu/signal.c | 13 +++++---
arch/x86/kernel/fpu/xstate.c | 55 --------------------------------------
3 files changed, 8 insertions(+), 61 deletions(-)
--- a/arch/x86/include/asm/fpu/xstate.h
+++ b/arch/x86/include/asm/fpu/xstate.h
@@ -107,7 +107,6 @@ struct membuf;
void copy_xstate_to_kernel(struct membuf to, struct xregs_state *xsave);
int copy_kernel_to_xstate(struct xregs_state *xsave, const void *kbuf);
int copy_user_to_xstate(struct xregs_state *xsave, const void __user *ubuf);
-void copy_supervisor_to_kernel(struct xregs_state *xsave);
void copy_dynamic_supervisor_to_kernel(struct xregs_state *xstate, u64 mask);
void copy_kernel_to_dynamic_supervisor(struct xregs_state *xstate, u64 mask);
--- a/arch/x86/kernel/fpu/signal.c
+++ b/arch/x86/kernel/fpu/signal.c
@@ -411,15 +411,18 @@ static int __fpu__restore_sig(void __use
* the optimisation).
*/
fpregs_lock();
-
if (!test_thread_flag(TIF_NEED_FPU_LOAD)) {
-
/*
- * Supervisor states are not modified by user space input. Save
- * current supervisor states first and invalidate the FPU regs.
+ * If supervisor states are available then save the
+ * hardware state in current's fpstate so that the
+ * supervisor state is preserved. Save the full state for
+ * simplicity. There is no point in optimizing this by only
+ * saving the supervisor states and then shuffle them to
+ * the right place in memory. This is the slow path and the
+ * above XRSTOR failed or ia32_fxstate is true. Shrug.
*/
if (xfeatures_mask_supervisor())
- copy_supervisor_to_kernel(&fpu->state.xsave);
+ copy_xregs_to_kernel(&fpu->state.xsave);
set_thread_flag(TIF_NEED_FPU_LOAD);
}
__fpu_invalidate_fpregs_state(fpu);
--- a/arch/x86/kernel/fpu/xstate.c
+++ b/arch/x86/kernel/fpu/xstate.c
@@ -1129,61 +1129,6 @@ int copy_user_to_xstate(struct xregs_sta
return 0;
}
-/*
- * Save only supervisor states to the kernel buffer. This blows away all
- * old states, and is intended to be used only in __fpu__restore_sig(), where
- * user states are restored from the user buffer.
- */
-void copy_supervisor_to_kernel(struct xregs_state *xstate)
-{
- struct xstate_header *header;
- u64 max_bit, min_bit;
- u32 lmask, hmask;
- int err, i;
-
- if (WARN_ON(!boot_cpu_has(X86_FEATURE_XSAVES)))
- return;
-
- if (!xfeatures_mask_supervisor())
- return;
-
- max_bit = __fls(xfeatures_mask_supervisor());
- min_bit = __ffs(xfeatures_mask_supervisor());
-
- lmask = xfeatures_mask_supervisor();
- hmask = xfeatures_mask_supervisor() >> 32;
- XSTATE_OP(XSAVES, xstate, lmask, hmask, err);
-
- /* We should never fault when copying to a kernel buffer: */
- if (WARN_ON_FPU(err))
- return;
-
- /*
- * At this point, the buffer has only supervisor states and must be
- * converted back to normal kernel format.
- */
- header = &xstate->header;
- header->xcomp_bv |= xfeatures_mask_all;
-
- /*
- * This only moves states up in the buffer. Start with
- * the last state and move backwards so that states are
- * not overwritten until after they are moved. Note:
- * memmove() allows overlapping src/dst buffers.
- */
- for (i = max_bit; i >= min_bit; i--) {
- u8 *xbuf = (u8 *)xstate;
-
- if (!((header->xfeatures >> i) & 1))
- continue;
-
- /* Move xfeature 'i' into its normal location */
- memmove(xbuf + xstate_comp_offsets[i],
- xbuf + xstate_supervisor_only_offsets[i],
- xstate_sizes[i]);
- }
-}
-
/**
* copy_dynamic_supervisor_to_kernel() - Save dynamic supervisor states to
* an xsave area
next prev parent reply other threads:[~2021-06-11 16:44 UTC|newest]
Thread overview: 69+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-11 16:15 [patch 00/41] x86/fpu: Spring cleaning and PKRU sanitizing Thomas Gleixner
2021-06-11 16:15 ` [patch 01/41] Revert a5eff7259790 ("x86/pkeys: Add PKRU value to init_fpstate") Thomas Gleixner
2021-06-11 17:04 ` Borislav Petkov
2021-06-11 16:15 ` [patch 02/41] x86/fpu: Mark various FPU states __ro_after_init Thomas Gleixner
2021-06-11 17:21 ` Borislav Petkov
2021-06-11 18:35 ` Andy Lutomirski
2021-06-11 16:15 ` [patch 03/41] x86/fpu: Remove unused get_xsave_field_ptr() Thomas Gleixner
2021-06-11 18:35 ` Andy Lutomirski
2021-06-11 16:15 ` [patch 04/41] x86/fpu: Move inlines where they belong Thomas Gleixner
2021-06-11 16:15 ` [patch 05/41] x86/fpu: Limit xstate copy size in xstateregs_set() Thomas Gleixner
2021-06-11 18:15 ` Borislav Petkov
2021-06-11 18:37 ` Andy Lutomirski
2021-06-11 19:37 ` Thomas Gleixner
2021-06-11 16:15 ` [patch 06/41] x86/fpu: Sanitize xstateregs_set() Thomas Gleixner
2021-06-11 18:45 ` Andy Lutomirski
2021-06-11 20:23 ` Thomas Gleixner
2021-06-11 16:15 ` [patch 07/41] x86/fpu: Simplify PTRACE_GETREGS code Thomas Gleixner
2021-06-11 18:47 ` Andy Lutomirski
2021-06-12 9:13 ` Borislav Petkov
2021-06-11 16:15 ` [patch 08/41] x86/fpu: Restrict fpstate sanitizing to legacy components Thomas Gleixner
2021-06-11 19:03 ` Andy Lutomirski
2021-06-11 19:18 ` Andy Lutomirski
2021-06-11 20:33 ` Thomas Gleixner
2021-06-11 20:34 ` Thomas Gleixner
2021-06-11 20:27 ` Thomas Gleixner
2021-06-11 22:12 ` Thomas Gleixner
2021-06-12 13:15 ` Thomas Gleixner
2021-06-12 22:05 ` Thomas Gleixner
2021-06-11 16:15 ` [patch 09/41] x86/kvm: Avoid looking up PKRU in XSAVE buffer Thomas Gleixner
2021-06-14 10:26 ` Borislav Petkov
2021-06-14 19:34 ` Dave Hansen
2021-06-15 10:09 ` Borislav Petkov
2021-06-11 16:15 ` [patch 10/41] x86/fpu: Cleanup arch_set_user_pkey_access() Thomas Gleixner
2021-06-11 16:15 ` Thomas Gleixner [this message]
2021-06-11 19:42 ` [patch 11/41] x86/fpu: Get rid of copy_supervisor_to_kernel() Andy Lutomirski
2021-06-11 16:15 ` [patch 12/41] x86/fpu: Rename copy_xregs_to_kernel() and copy_kernel_to_xregs() Thomas Gleixner
2021-06-11 16:15 ` [patch 13/41] x86/fpu: Rename copy_user_to_xregs() and copy_xregs_to_user() Thomas Gleixner
2021-06-11 16:15 ` [patch 14/41] x86/fpu: Rename fxregs related copy functions Thomas Gleixner
2021-06-11 16:15 ` [patch 15/41] x86/fpu: Rename fregs " Thomas Gleixner
2021-06-11 16:15 ` [patch 16/41] x86/fpu: Rename xstate copy functions which are related to UABI Thomas Gleixner
2021-06-11 16:15 ` [patch 17/41] x86/fpu: Deduplicate copy_uabi_from_user/kernel_to_xstate() Thomas Gleixner
2021-06-11 16:15 ` [patch 18/41] x86/fpu: Rename copy_fpregs_to_fpstate() to save_fpregs_to_fpstate() Thomas Gleixner
2021-06-11 16:15 ` [patch 19/41] x86/fpu: Rename copy_kernel_to_fpregs() to restore_fpregs_from_kernel() Thomas Gleixner
2021-06-11 16:15 ` [patch 20/41] x86/fpu: Rename initstate copy functions Thomas Gleixner
2021-06-11 16:15 ` [patch 21/41] x86/fpu: Rename "dynamic" XSTATEs to "independent" Thomas Gleixner
2021-06-11 16:15 ` [patch 22/41] x86/fpu/xstate: Sanitize handling of independent features Thomas Gleixner
2021-06-11 16:15 ` [patch 23/41] x86/pkeys: Move read_pkru() and write_pkru() Thomas Gleixner
2021-06-11 16:15 ` [patch 24/41] x86/fpu: Differentiate "copy" versus "move" of fpregs Thomas Gleixner
2021-06-11 16:15 ` [patch 25/41] x86/cpu: Sanitize X86_FEATURE_OSPKE Thomas Gleixner
2021-06-11 16:15 ` [patch 26/41] x86/pkru: Provide pkru_get_init_value() Thomas Gleixner
2021-06-11 16:15 ` [patch 27/41] x86/pkru: Provide pkru_write_default() Thomas Gleixner
2021-06-11 16:15 ` [patch 28/41] x86/cpu: Write the default PKRU value when enabling PKE Thomas Gleixner
2021-06-11 16:15 ` [patch 29/41] x86/fpu: Use pkru_write_default() in copy_init_fpstate_to_fpregs() Thomas Gleixner
2021-06-11 16:15 ` [patch 30/41] x86/fpu: Rename fpu__clear_all() to fpu_flush_thread() Thomas Gleixner
2021-06-11 16:15 ` [patch 31/41] x86/fpu: Clean up the fpu__clear() variants Thomas Gleixner
2021-06-11 16:15 ` [patch 32/41] x86/fpu: Rename __fpregs_load_activate() to fpregs_restore_userregs() Thomas Gleixner
2021-06-11 16:15 ` [patch 33/41] x86/fpu: Move FXSAVE_LEAK quirk info __copy_kernel_to_fpregs() Thomas Gleixner
2021-06-11 16:15 ` [patch 34/41] x86/fpu: Rename xfeatures_mask_user() to xfeatures_mask_uabi() Thomas Gleixner
2021-06-11 16:15 ` [patch 35/41] x86/fpu: Dont restore PKRU in fpregs_restore_userspace() Thomas Gleixner
2021-06-11 16:15 ` [patch 36/41] x86/fpu: Add PKRU storage outside of task XSAVE buffer Thomas Gleixner
2021-06-11 16:16 ` [patch 37/41] x86/fpu: Hook up PKRU into ptrace() Thomas Gleixner
2021-06-11 16:16 ` [patch 38/41] x86/fpu: Mask PKRU from kernel XRSTOR[S] operations Thomas Gleixner
2021-06-11 16:16 ` [patch 39/41] x86/fpu: Remove PKRU handling from switch_fpu_finish() Thomas Gleixner
2021-06-11 16:16 ` [patch 40/41] x86/fpu: Dont store PKRU in xstate in fpu_reset_fpstate() Thomas Gleixner
2021-06-11 16:16 ` [patch 41/41] x86/pkru: Remove xstate fiddling from write_pkru() Thomas Gleixner
2021-06-12 0:24 ` [patch 00/41] x86/fpu: Spring cleaning and PKRU sanitizing Thomas Gleixner
2021-06-12 0:40 ` Dave Hansen
2021-06-16 20:55 ` Dave Hansen
2021-06-17 7:06 ` Thomas Gleixner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210611163112.153392869@linutronix.de \
--to=tglx@linutronix.de \
--cc=bigeasy@linutronix.de \
--cc=bp@suse.de \
--cc=dave.hansen@linux.intel.com \
--cc=fenghua.yu@intel.com \
--cc=kan.liang@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=luto@kernel.org \
--cc=peterz@infradead.org \
--cc=tony.luck@intel.com \
--cc=yu-cheng.yu@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.