[PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Lukas Wunner]

Andrew reports the following build breakage of arm allmodconfig,
reproducible with gcc 14.2.0 and 15.2.0:

  crypto/ecc.c: In function 'ecc_point_mult':
  crypto/ecc.c:1380:1: error: the frame size of 1360 bytes is larger than 1280 bytes [-Werror=frame-larger-than=]

gcc excessively inlines functions called by ecc_point_mult() (without
there being any explicit inline declarations) and doesn't seem smart
enough to stay below CONFIG_FRAME_WARN.

clang does not exhibit the issue.

The issue only occurs with CONFIG_KASAN_STACK=y because it enlarges the
frame size.  This has been a controversial topic a couple of times:

https://lore.kernel.org/r/CAK8P3a3_Tdc-XVPXrJ69j3S9048uzmVJGrNcvi0T6yr6OrHkPw@mail.gmail.com/

Prevent gcc from going overboard with inlining to unbreak the build.
The maximum inline limit to avoid the error is 101.  Use 100 to get a
nice round number per Andrew's preference.

Reported-by: Andrew Morton <akpm@linux-foundation.org> # off-list
Signed-off-by: Lukas Wunner <lukas@wunner.de>
---
 crypto/Makefile | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/crypto/Makefile b/crypto/Makefile
index 04e269117589..b3ac7f29153e 100644
--- a/crypto/Makefile
+++ b/crypto/Makefile
@@ -181,6 +181,11 @@ obj-$(CONFIG_CRYPTO_ZSTD) += zstd.o
 obj-$(CONFIG_CRYPTO_ECC) += ecc.o
 obj-$(CONFIG_CRYPTO_ESSIV) += essiv.o
 
+# Avoid exceeding stack frame due to excessive gcc inlining in ecc_point_mult()
+ifeq ($(ARCH)$(CONFIG_KASAN_STACK)$(LLVM),army)
+CFLAGS_ecc.o += $(call cc-option,-finline-limit=100)
+endif
+
 ecdh_generic-y += ecdh.o
 ecdh_generic-y += ecdh_helper.o
 obj-$(CONFIG_CRYPTO_ECDH) += ecdh_generic.o
-- 
2.51.0

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Andy Shevchenko]

On Wed, Apr 08, 2026 at 08:15:49AM +0200, Lukas Wunner wrote:
> Andrew reports the following build breakage of arm allmodconfig,
> reproducible with gcc 14.2.0 and 15.2.0:
> 
>   crypto/ecc.c: In function 'ecc_point_mult':
>   crypto/ecc.c:1380:1: error: the frame size of 1360 bytes is larger than 1280 bytes [-Werror=frame-larger-than=]
> 
> gcc excessively inlines functions called by ecc_point_mult() (without
> there being any explicit inline declarations) and doesn't seem smart
> enough to stay below CONFIG_FRAME_WARN.
> 
> clang does not exhibit the issue.
> 
> The issue only occurs with CONFIG_KASAN_STACK=y because it enlarges the
> frame size.  This has been a controversial topic a couple of times:
> 
> https://lore.kernel.org/r/CAK8P3a3_Tdc-XVPXrJ69j3S9048uzmVJGrNcvi0T6yr6OrHkPw@mail.gmail.com/
> 
> Prevent gcc from going overboard with inlining to unbreak the build.
> The maximum inline limit to avoid the error is 101.  Use 100 to get a
> nice round number per Andrew's preference.

I think this is not the best solution. We still can refactor the code and avoid
being dependant to the (useful) kernel options.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Lukas Wunner]

On Wed, Apr 08, 2026 at 02:31:21PM +0300, Andy Shevchenko wrote:
> On Wed, Apr 08, 2026 at 08:15:49AM +0200, Lukas Wunner wrote:
> > Andrew reports the following build breakage of arm allmodconfig,
> > reproducible with gcc 14.2.0 and 15.2.0:
> > 
> >   crypto/ecc.c: In function 'ecc_point_mult':
> >   crypto/ecc.c:1380:1: error: the frame size of 1360 bytes is larger than 1280 bytes [-Werror=frame-larger-than=]
> > 
> > gcc excessively inlines functions called by ecc_point_mult() (without
> > there being any explicit inline declarations) and doesn't seem smart
> > enough to stay below CONFIG_FRAME_WARN.
> > 
> > clang does not exhibit the issue.
> > 
> > The issue only occurs with CONFIG_KASAN_STACK=y because it enlarges the
> > frame size.  This has been a controversial topic a couple of times:
> > 
> > https://lore.kernel.org/r/CAK8P3a3_Tdc-XVPXrJ69j3S9048uzmVJGrNcvi0T6yr6OrHkPw@mail.gmail.com/
> > 
> > Prevent gcc from going overboard with inlining to unbreak the build.
> > The maximum inline limit to avoid the error is 101.  Use 100 to get a
> > nice round number per Andrew's preference.
> 
> I think this is not the best solution. We still can refactor the code
> and avoid being dependant to the (useful) kernel options.

Refactor how?  Mark functions "noinline"?  That may negatively impact
performance for everyone.

Note that this is a different kind of stack frame exhaustion than the one
in drivers/mtd/chips/cfi_cmdset_0001.c:do_write_buffer():  The latter
is a single function with lots of large local variables, whereas
ecc_point_mult() itself has a reasonable number of variables on the stack,
but gcc inlines numerous function calls that each increase the stack frame.

And gcc isn't smart enough to stop inlining when it reaches the maximum
stack frame size allowed by CONFIG_FRAME_WARN.

It's apparently a compiler bug.  Why should we work around compiler bugs
by refactoring the code?  The proposed patch instructs gcc to limit
inlining and we can easily remove that once the bug is fixed.

As Arnd explains in the above-linked message, stack frame exhaustion
in crypto/ tends to be caused by compiler bugs.  There are already two
other workarounds for compiler bugs in crypto/Makefile, one for wp512.o
and another for serpent_generic.o.  Amending CFLAGS is how we've dealt
with these issues in the past, not by refactoring code.

Thanks,

Lukas

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Andy Shevchenko]

On Wed, Apr 08, 2026 at 03:36:47PM +0200, Lukas Wunner wrote:
> On Wed, Apr 08, 2026 at 02:31:21PM +0300, Andy Shevchenko wrote:
> > On Wed, Apr 08, 2026 at 08:15:49AM +0200, Lukas Wunner wrote:
> > > Andrew reports the following build breakage of arm allmodconfig,
> > > reproducible with gcc 14.2.0 and 15.2.0:
> > > 
> > >   crypto/ecc.c: In function 'ecc_point_mult':
> > >   crypto/ecc.c:1380:1: error: the frame size of 1360 bytes is larger than 1280 bytes [-Werror=frame-larger-than=]
> > > 
> > > gcc excessively inlines functions called by ecc_point_mult() (without
> > > there being any explicit inline declarations) and doesn't seem smart
> > > enough to stay below CONFIG_FRAME_WARN.
> > > 
> > > clang does not exhibit the issue.
> > > 
> > > The issue only occurs with CONFIG_KASAN_STACK=y because it enlarges the
> > > frame size.  This has been a controversial topic a couple of times:
> > > 
> > > https://lore.kernel.org/r/CAK8P3a3_Tdc-XVPXrJ69j3S9048uzmVJGrNcvi0T6yr6OrHkPw@mail.gmail.com/
> > > 
> > > Prevent gcc from going overboard with inlining to unbreak the build.
> > > The maximum inline limit to avoid the error is 101.  Use 100 to get a
> > > nice round number per Andrew's preference.
> > 
> > I think this is not the best solution. We still can refactor the code
> > and avoid being dependant to the (useful) kernel options.

> Refactor how?  Mark functions "noinline"?  That may negatively impact
> performance for everyone.
> 
> Note that this is a different kind of stack frame exhaustion than the one
> in drivers/mtd/chips/cfi_cmdset_0001.c:do_write_buffer():  The latter
> is a single function with lots of large local variables, whereas
> ecc_point_mult() itself has a reasonable number of variables on the stack,
> but gcc inlines numerous function calls that each increase the stack frame.

Ah, that makes the difference, thanks for elaborating!

> And gcc isn't smart enough to stop inlining when it reaches the maximum
> stack frame size allowed by CONFIG_FRAME_WARN.
> 
> It's apparently a compiler bug.  Why should we work around compiler bugs
> by refactoring the code?  The proposed patch instructs gcc to limit
> inlining and we can easily remove that once the bug is fixed.
> 
> As Arnd explains in the above-linked message, stack frame exhaustion
> in crypto/ tends to be caused by compiler bugs.  There are already two
> other workarounds for compiler bugs in crypto/Makefile, one for wp512.o
> and another for serpent_generic.o.  Amending CFLAGS is how we've dealt
> with these issues in the past, not by refactoring code.

Yeah, that's the way we may deal with the issue.

Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Nathan Chancellor]

On Wed, Apr 08, 2026 at 08:15:49AM +0200, Lukas Wunner wrote:
> Andrew reports the following build breakage of arm allmodconfig,
> reproducible with gcc 14.2.0 and 15.2.0:
> 
>   crypto/ecc.c: In function 'ecc_point_mult':
>   crypto/ecc.c:1380:1: error: the frame size of 1360 bytes is larger than 1280 bytes [-Werror=frame-larger-than=]
> 
> gcc excessively inlines functions called by ecc_point_mult() (without
> there being any explicit inline declarations) and doesn't seem smart
> enough to stay below CONFIG_FRAME_WARN.
> 
> clang does not exhibit the issue.
> 
> The issue only occurs with CONFIG_KASAN_STACK=y because it enlarges the
> frame size.  This has been a controversial topic a couple of times:
> 
> https://lore.kernel.org/r/CAK8P3a3_Tdc-XVPXrJ69j3S9048uzmVJGrNcvi0T6yr6OrHkPw@mail.gmail.com/
> 
> Prevent gcc from going overboard with inlining to unbreak the build.
> The maximum inline limit to avoid the error is 101.  Use 100 to get a
> nice round number per Andrew's preference.
> 
> Reported-by: Andrew Morton <akpm@linux-foundation.org> # off-list
> Signed-off-by: Lukas Wunner <lukas@wunner.de>
> ---
>  crypto/Makefile | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/crypto/Makefile b/crypto/Makefile
> index 04e269117589..b3ac7f29153e 100644
> --- a/crypto/Makefile
> +++ b/crypto/Makefile
> @@ -181,6 +181,11 @@ obj-$(CONFIG_CRYPTO_ZSTD) += zstd.o
>  obj-$(CONFIG_CRYPTO_ECC) += ecc.o
>  obj-$(CONFIG_CRYPTO_ESSIV) += essiv.o
>  
> +# Avoid exceeding stack frame due to excessive gcc inlining in ecc_point_mult()
> +ifeq ($(ARCH)$(CONFIG_KASAN_STACK)$(LLVM),army)

Please use proper Kconfig variables here.

  ifeq ($(CONFIG_ARM)$(CONFIG_KASAN_STACK)$(CONFIG_CC_IS_GCC),yyy)

Which is both more robust, as $(LLVM) may not be set but CC=clang could
be, and it is clearer (in my opinion). If all supported versions of GCC
support this flag, you could drop the cc-option at that point.

> +CFLAGS_ecc.o += $(call cc-option,-finline-limit=100)
> +endif
> +
>  ecdh_generic-y += ecdh.o
>  ecdh_generic-y += ecdh_helper.o
>  obj-$(CONFIG_CRYPTO_ECDH) += ecdh_generic.o
> -- 
> 2.51.0
> 

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Arnd Bergmann]

On Wed, Apr 8, 2026, at 15:36, Lukas Wunner wrote:
> On Wed, Apr 08, 2026 at 02:31:21PM +0300, Andy Shevchenko wrote:
>> On Wed, Apr 08, 2026 at 08:15:49AM +0200, Lukas Wunner wrote:
>> > Prevent gcc from going overboard with inlining to unbreak the build.
>> > The maximum inline limit to avoid the error is 101.  Use 100 to get a
>> > nice round number per Andrew's preference.

Have you checked if the total call chain gets a lower stack usage this
way? Usually the high stack usage is a sign of absolutely awful
code generation when the compiler runs into a corner case that
spills variables onto the stack instead of keeping them in registers.

The question is whether the lower inline limit causes the compiler
to not get into this state at all and produce the expected object
code, or if it just ends up producing multiple functions that
stay under the limit individually but have the same problems with
stack usage and performance as before.

I think your patch can be merged either way, but it would be
good to describe what type of problem we are hitting here.
 
>> I think this is not the best solution. We still can refactor the code
>> and avoid being dependant to the (useful) kernel options.
>
> Refactor how?  Mark functions "noinline"?  That may negatively impact
> performance for everyone.

I ran into the same issue last year and worked around it by
turning off kasan for this file, which of course is problematic
for other reasons, and I never submitted my hack for inclusion:

--- a/crypto/Makefile
+++ b/crypto/Makefile
@@ -176,6 +176,7 @@ obj-$(CONFIG_CRYPTO_USER_API_RNG) += algif_rng.o
 obj-$(CONFIG_CRYPTO_USER_API_AEAD) += algif_aead.o
 obj-$(CONFIG_CRYPTO_ZSTD) += zstd.o
 obj-$(CONFIG_CRYPTO_ECC) += ecc.o
+KASAN_SANITIZE_ecc.o = n
 obj-$(CONFIG_CRYPTO_ESSIV) += essiv.o
 
 ecdh_generic-y += ecdh.o

In principle this could be done on a per-function basis.

      Arnd

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Lukas Wunner]

[-- Attachment #1: Type: text/plain, Size: 1950 bytes --]

On Mon, Apr 13, 2026 at 05:42:39PM +0200, Arnd Bergmann wrote:
> On Wed, Apr 8, 2026, at 15:36, Lukas Wunner wrote:
> > On Wed, Apr 08, 2026 at 02:31:21PM +0300, Andy Shevchenko wrote:
> > > On Wed, Apr 08, 2026 at 08:15:49AM +0200, Lukas Wunner wrote:
> > > > Prevent gcc from going overboard with inlining to unbreak the build.
> > > > The maximum inline limit to avoid the error is 101.  Use 100 to get a
> > > > nice round number per Andrew's preference.
> 
> Have you checked if the total call chain gets a lower stack usage this
> way? Usually the high stack usage is a sign of absolutely awful
> code generation when the compiler runs into a corner case that
> spills variables onto the stack instead of keeping them in registers.
> 
> The question is whether the lower inline limit causes the compiler
> to not get into this state at all and produce the expected object
> code, or if it just ends up producing multiple functions that
> stay under the limit individually but have the same problems with
> stack usage and performance as before.

Attached please find the Assembler output created by gcc -save-temps,
both the original version and the one with limited inlining.

The former requires a 1360 bytes stack frame, the latter 1232 bytes.
E.g. xycz_initial_double() is not inlined into ecc_point_mult(),
together with all its recursive baggage, so the latter version
contains two branch instructions to that function which the former
(original) version does not contain.

At the beginning of the function, it looks like the same register values
are stored to multiple locations on the stack.  I assume that's what you
mean by awful code generation?  This odd behavior seems more subdued in
the version with limited inlining.

> I think your patch can be merged either way, but it would be
> good to describe what type of problem we are hitting here.

I will respin and I will also take Nathan's suggestion into account.

Thanks,

Lukas

[-- Attachment #2: ecc_point_mult_orig.s --]
[-- Type: text/plain, Size: 23719 bytes --]

	.section	.rodata.ecc_point_mult.str1.4,"aMS",%progbits,1
	.align	2
.LC43:
	.ascii	"7 48 72 6 z:1337 160 72 6 z:1208 272 72 7 t1:1196 3"
	.ascii	"84 144 7 rx:1335 592 144 7 ry:1336 800 144 7 sk:133"
	.ascii	"8 1008 144 12 product:1018\000"
	.section	.text.ecc_point_mult,"ax",%progbits
	.align	2
	.syntax unified
	.arm
	.type	ecc_point_mult, %function
ecc_point_mult:
.LASANPC2762:
	.fnstart
	@ args = 8, pretend = 0, frame = 1360
	@ frame_needed = 0, uses_anonymous_args = 0
	push	{r4, r5, r6, r7, r8, r9, r10, fp, lr}
	.save {r4, r5, r6, r7, r8, r9, r10, fp, lr}
	.pad #1360
	sub	sp, sp, #1360
	.pad #12
	sub	sp, sp, #12
	push	{lr}
	bl	__gnu_mcount_nc
	str	r3, [sp, #80]
	add	r3, sp, #1360
	add	r3, r3, #8
	bic	r9, r3, #31
	sub	r3, r9, #1248
	ldr	ip, .L2798
	lsr	r3, r3, #3
	add	r7, r3, #-1627389952
	str	r0, [sp, #52]
	str	r1, [sp, #32]
	ldr	r4, [sp, #1408]
	str	r9, [sp, #44]
	str	r7, [sp, #68]
	str	ip, [r9, #-1248]
	ldr	ip, .L2798+4
	ldr	r3, .L2798+8
	str	ip, [r9, #-1244]
	ldr	ip, .L2798+12
	ldr	lr, .L2798+16
	ldr	r0, .L2798+20
	ldr	r1, .L2798+24
	ldr	r5, .L2798+28
	ldr	r6, .L2798+32
	mov	r8, r2
	ldr	r2, .L2798+36
	str	ip, [r9, #-1240]
	mov	ip, #-234881024
	str	r3, [r7, #16]
	str	r3, [r7, #44]
	str	r3, [r7, #68]
	str	r3, [r7, #92]
	str	r3, [r7, #96]
	str	r3, [r7, #120]
	ldr	r3, .L2798+40
	str	lr, [r7, #28]
	str	ip, [r7, #12]
	str	ip, [r7, #40]
	str	r5, [r7, #4]
	str	r2, [r7, #32]
	str	r0, [r7, #64]
	str	r0, [r7, #116]
	str	r2, [r7, #72]
	str	r2, [r7, #124]
	str	r1, [r7, #144]
	str	r1, [r7, #148]
	str	r6, [r7]
	ldr	r3, [r3]
	str	r3, [sp, #1364]
	mov	r3, #0
	bl	__sanitizer_cov_trace_pc
	sub	r3, r9, #864
	mov	r2, #144
	mov	r1, #254
	mov	r0, r3
	sub	r5, r9, #448
	str	r3, [sp, #40]
	bl	memset
	sub	r3, r9, #656
	mov	r2, #144
	mov	r1, #254
	mov	r0, r3
	str	r3, [sp, #24]
	bl	memset
	sub	r3, r9, #1200
	mov	r0, r3
	mov	r2, #72
	mov	r1, #254
	str	r3, [sp, #36]
	bl	memset
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	add	r0, r4, #20
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2534
	bl	__asan_report_load4_noabort
.L2534:
	add	r0, r4, #24
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	ldr	r3, [r4, #20]
	str	r3, [sp, #56]
	beq	.L2535
	bl	__asan_report_load4_noabort
.L2535:
	ldr	r1, [sp, #1412]
	mov	r0, #0
	ldr	r7, [r4, #24]
	bl	__sanitizer_cov_trace_const_cmp4
	ldr	r3, [sp, #1412]
	cmp	r3, #0
	beq	.L2536
	mov	r6, r5
	mov	r1, #0
	mov	r10, r7
	strd	r6, [sp, #16]
	mov	r9, r5
	mov	r7, r1
	str	r1, [sp, #8]
	str	r4, [sp, #28]
	str	r5, [sp, #48]
.L2541:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r8, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2537
	mov	r0, r8
	bl	__asan_report_load8_noabort
.L2537:
	lsr	r3, r10, #3
	add	r3, r3, #-1627389952
	ldm	r8, {r6, fp}
	ldrsb	r3, [r3]
	add	r8, r8, #8
	mov	r5, r10
	cmp	r3, #0
	beq	.L2538
	mov	r0, r10
	bl	__asan_report_load8_noabort
.L2538:
	ldrd	r4, [r5]
	ldr	r3, [sp, #8]
	mov	r0, r6
	adds	r4, r6, r4
	adc	r5, fp, r5
	adds	r4, r4, r3
	adc	r5, r5, #0
	mov	r2, r4
	mov	r3, r5
	mov	r1, fp
	bl	__sanitizer_cov_trace_cmp8
	cmp	fp, r5
	cmpeq	r6, r4
	add	r10, r10, #8
	beq	.L2539
	bl	__sanitizer_cov_trace_pc
	mov	r3, r5
	mov	r1, fp
	mov	r2, r4
	mov	r0, r6
	bl	__sanitizer_cov_trace_cmp8
	cmp	r4, r6
	sbcs	r1, r5, fp
	movcc	r3, #1
	movcs	r3, #0
	str	r3, [sp, #8]
.L2539:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r9, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2540
	mov	r0, r9
	bl	__asan_report_store8_noabort
.L2540:
	add	r7, r7, #1
	strd	r4, [r9]
	ldr	r0, [sp, #1412]
	mov	r1, r7
	bl	__sanitizer_cov_trace_cmp4
	add	r9, r9, #8
	ldr	r3, [sp, #1412]
	cmp	r7, r3
	bne	.L2541
	ldr	r4, [sp, #28]
	ldr	r5, [sp, #48]
	ldrd	r6, [sp, #16]
	bl	__sanitizer_cov_trace_pc
	mov	r10, #0
	str	r10, [sp, #16]
	str	r4, [sp, #20]
	str	r5, [sp, #28]
	b	.L2550
.L2536:
	bl	__sanitizer_cov_trace_pc
	add	r0, r4, #4
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2543
	bl	__asan_report_load4_noabort
.L2543:
	ldr	r8, [r4, #4]
	ldr	r5, .L2798+44
	mov	r1, r8
	mov	r0, r5
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r8, r5
	ldr	r3, [sp, #44]
	beq	.L2632
	sub	r3, r3, #376
	str	r3, [sp, #28]
	b	.L2545
.L2633:
	mov	r6, r9
.L2550:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2546
	mov	r0, r6
	bl	__asan_report_load8_noabort
.L2546:
	lsr	r3, r7, #3
	add	r3, r3, #-1627389952
	ldm	r6, {r8, fp}
	ldrsb	r3, [r3]
	add	r9, r6, #8
	mov	r5, r7
	cmp	r3, #0
	beq	.L2547
	mov	r0, r7
	bl	__asan_report_load8_noabort
.L2547:
	ldrd	r4, [r5]
	ldr	r3, [sp, #16]
	mov	r0, r8
	adds	r4, r8, r4
	adc	r5, fp, r5
	adds	r4, r4, r3
	adc	r5, r5, #0
	mov	r2, r4
	mov	r3, r5
	mov	r1, fp
	bl	__sanitizer_cov_trace_cmp8
	cmp	fp, r5
	cmpeq	r8, r4
	add	r7, r7, #8
	beq	.L2548
	bl	__sanitizer_cov_trace_pc
	mov	r3, r5
	mov	r1, fp
	mov	r2, r4
	mov	r0, r8
	bl	__sanitizer_cov_trace_cmp8
	cmp	r4, r8
	sbcs	r1, r5, fp
	movcc	r3, #1
	movcs	r3, #0
	str	r3, [sp, #16]
.L2548:
	add	r6, r6, #72
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2549
	mov	r0, r6
	bl	__asan_report_store8_noabort
.L2549:
	add	r10, r10, #1
	ldr	r0, [sp, #1412]
	mov	r1, r10
	strd	r4, [r9, #64]
	bl	__sanitizer_cov_trace_cmp4
	ldr	r3, [sp, #1412]
	cmp	r10, r3
	bne	.L2633
	ldr	r4, [sp, #20]
	ldr	r5, [sp, #28]
	bl	__sanitizer_cov_trace_pc
	add	r0, r4, #4
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	ldrsb	r2, [r2]
	ldr	r3, [sp, #8]
	eor	r3, r3, #1
	rsb	r3, r3, #0
	and	r3, r3, #72
	add	r3, r5, r3
	str	r3, [sp, #28]
	and	r3, r0, #7
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2552
	bl	__asan_report_load4_noabort
.L2552:
	ldr	r6, [r4, #4]
	ldr	r5, .L2798+44
	mov	r1, r6
	mov	r0, r5
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r6, r5
	beq	.L2553
.L2545:
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #1412]
	lsl	r8, r3, #6
	bl	__sanitizer_cov_trace_pc
	sub	r8, r8, #1
	ldr	r1, [sp, #32]
	lsr	r3, r1, #3
	str	r3, [sp, #76]
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	str	r3, [sp, #72]
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2554
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L2554:
	ldr	r3, [sp, #32]
	ldr	r1, [sp, #1412]
	mov	r0, #0
	ldr	r5, [r3]
	bl	__sanitizer_cov_trace_const_cmp4
	ldr	r3, [sp, #1412]
	cmp	r3, #0
	bne	.L2630
	ldr	r3, [sp, #44]
	sub	r2, r3, #792
	str	r2, [sp, #48]
	ldr	r2, [sp, #32]
	sub	r3, r3, #584
	add	r2, r2, #4
	str	r2, [sp, #64]
	str	r3, [sp, #60]
	b	.L2559
.L2630:
	ldr	r3, [sp, #44]
	ldr	r9, [sp, #1412]
	sub	r3, r3, #792
	mov	r6, r3
	mov	r7, #0
	str	r3, [sp, #48]
.L2557:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r5, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2555
	mov	r0, r5
	bl	__asan_report_load8_noabort
.L2555:
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrd	r10, [r5], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2556
	mov	r0, r6
	bl	__asan_report_store8_noabort
.L2556:
	add	r7, r7, #1
	mov	r1, r7
	mov	r0, r9
	strd	r10, [r6], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r7, r9
	bne	.L2557
	b	.L2795
.L2632:
	sub	r2, r3, #376
	str	r2, [sp, #28]
	ldr	r2, [sp, #32]
	lsr	r1, r2, #3
	str	r1, [sp, #76]
	and	r1, r2, #7
	str	r1, [sp, #72]
	add	r2, r2, #4
	sub	r1, r3, #792
	sub	r3, r3, #584
	str	r1, [sp, #48]
	str	r2, [sp, #64]
	str	r3, [sp, #60]
	b	.L2559
.L2795:
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #32]
	add	r2, r3, #4
	and	r3, r2, #7
	str	r2, [sp, #64]
	lsr	r2, r2, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2560
	ldr	r0, [sp, #64]
	bl	__asan_report_load4_noabort
.L2560:
	ldr	r3, [sp, #44]
	ldr	r9, [sp, #1412]
	sub	r5, r3, #584
	ldr	r3, [sp, #32]
	mov	r7, #0
	str	r5, [sp, #60]
	ldr	r6, [r3, #4]
.L2563:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2561
	mov	r0, r6
	bl	__asan_report_load8_noabort
.L2561:
	lsr	r3, r5, #3
	add	r3, r3, #-1627389952
	ldrd	r10, [r6], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2562
	mov	r0, r5
	bl	__asan_report_store8_noabort
.L2562:
	add	r7, r7, #1
	mov	r1, r7
	mov	r0, r9
	strd	r10, [r5], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r7, r9
	bne	.L2563
.L2559:
	bl	__sanitizer_cov_trace_pc
	mov	r2, #72
	mov	r1, #254
	ldr	r3, [sp, #44]
	sub	fp, r3, #1088
	mov	r0, fp
	bl	memset
	add	r1, r4, #16
	and	r3, r1, #7
	lsr	r2, r1, #3
	str	r2, [sp, #20]
	add	r2, r2, #-1627389952
	str	r1, [sp, #84]
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	and	r3, r1, #7
	str	r3, [sp, #16]
	beq	.L2564
	mov	r0, r1
	bl	__asan_report_load1_noabort
.L2564:
	ldrb	r5, [r4, #16]	@ zero_extendqisi2
	mov	r0, #0
	mov	r1, r5
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r5, #0
	beq	.L2565
	ldr	r7, [sp, #48]
	mov	r6, #0
	mov	r9, fp
.L2568:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r7, #3
	add	r3, r3, #-1627389952
	sub	r2, r7, #72
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2566
	mov	r0, r7
	str	r2, [sp, #8]
	bl	__asan_report_load8_noabort
	ldr	r2, [sp, #8]
.L2566:
	lsr	r3, r2, #3
	add	r3, r3, #-1627389952
	ldrd	r10, [r7], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2567
	mov	r0, r2
	bl	__asan_report_store8_noabort
.L2567:
	add	r6, r6, #1
	mov	r1, r6
	mov	r0, r5
	strd	r10, [r7, #-80]
	bl	__sanitizer_cov_trace_cmp4
	cmp	r6, r5
	bne	.L2568
	ldr	r7, [sp, #60]
	mov	r6, #0
.L2571:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r7, #3
	add	r3, r3, #-1627389952
	sub	r2, r7, #72
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2569
	mov	r0, r7
	str	r2, [sp, #8]
	bl	__asan_report_load8_noabort
	ldr	r2, [sp, #8]
.L2569:
	lsr	r3, r2, #3
	add	r3, r3, #-1627389952
	ldrd	r10, [r7], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2570
	mov	r0, r2
	bl	__asan_report_store8_noabort
.L2570:
	add	r6, r6, #1
	mov	r1, r6
	mov	r0, r5
	strd	r10, [r7, #-80]
	bl	__sanitizer_cov_trace_cmp4
	cmp	r6, r5
	bne	.L2571
	mov	fp, r9
	mov	r10, r9
	mov	r6, #0
.L2573:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r9, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2572
	mov	r0, r9
	bl	__asan_report_store8_noabort
.L2572:
	mov	r2, #0
	mov	r3, #0
	add	r6, r6, #1
	mov	r1, r6
	mov	r0, r5
	strd	r2, [r9], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r6, r5
	bne	.L2573
	b	.L2796
.L2565:
	bl	__sanitizer_cov_trace_pc
	mov	r2, #1
	mov	r3, #0
	strd	r2, [fp]
	b	.L2575
.L2796:
	bl	__sanitizer_cov_trace_pc
	mov	r2, #1
	mov	r3, #0
	strd	r2, [fp]
	ldr	r6, [sp, #80]
	cmp	r6, #0
	beq	.L2575
	mov	r7, #0
.L2579:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2577
	mov	r0, r6
	bl	__asan_report_load8_noabort
.L2577:
	lsr	r3, r10, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	ldrd	r2, [r6], #8
	beq	.L2578
	mov	r0, r10
	strd	r2, [sp, #8]
	bl	__asan_report_store8_noabort
	ldrd	r2, [sp, #8]
.L2578:
	add	r7, r7, #1
	mov	r1, r7
	mov	r0, r5
	strd	r2, [r10], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r7, r5
	bne	.L2579
.L2575:
	bl	__sanitizer_cov_trace_pc
	mov	r2, #72
	mov	r1, #254
	ldr	r5, [sp, #44]
	sub	r9, r5, #976
	mov	r0, r9
	bl	memset
	mov	r2, r4
	mov	r1, fp
	mov	r0, r9
	bl	vli_mod_square_fast
	sub	r3, r5, #240
	mov	r2, #144
	mov	r0, r3
	mov	r1, #254
	str	r3, [sp, #8]
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2581
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2581:
	ldr	r5, [sp, #8]
	ldr	r6, [sp, #48]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r9
	mov	r1, r6
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r6
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2582
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2582:
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, fp
	mov	r1, r9
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r9
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2583
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2583:
	ldr	r5, [sp, #60]
	ldr	r6, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r9
	mov	r1, r5
	mov	r0, r6
	bl	vli_mult
	mov	r2, r4
	mov	r1, r6
	mov	r0, r5
	bl	vli_mmod_fast
	mov	r3, r4
	mov	r2, fp
	mov	r1, r5
	ldr	r0, [sp, #48]
	bl	ecc_point_double_jacobian
	mov	r2, #72
	mov	r1, #254
	mov	r0, r9
	bl	memset
	mov	r2, r4
	mov	r1, fp
	mov	r0, r9
	bl	vli_mod_square_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2584
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2584:
	ldr	r5, [sp, #8]
	ldr	r6, [sp, #40]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r9
	mov	r1, r6
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r6
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2585
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2585:
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, fp
	mov	r1, r9
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r9
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2586
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2586:
	ldr	r6, [sp, #24]
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r9
	mov	r1, r6
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r6
	bl	vli_mmod_fast
	mov	r1, r8
	mov	r0, #0
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r8, #0
	ble	.L2621
	ldr	fp, [sp, #40]
	mov	r7, #1
	str	r9, [sp, #44]
	b	.L2587
.L2799:
	.align	2
.L2798:
	.word	1102416563
	.word	.LC43
	.word	-218959118
	.word	.LASANPC2762
	.word	-218959360
	.word	-219021312
	.word	-202116109
	.word	61937
	.word	-235802127
	.word	62194
	.word	__stack_chk_guard
	.word	521
.L2621:
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #28]
	lsr	r3, r3, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2588
	ldr	r0, [sp, #28]
	bl	__asan_report_load8_noabort
.L2588:
	ldr	r3, [sp, #28]
	ldr	r5, [sp, #40]
	ldr	r1, [sp, #24]
	ldr	r3, [r3]
	str	r4, [sp]
	and	r3, r3, #1
	eor	r2, r3, #1
	rsb	r2, r2, #0
	rsb	r3, r3, #0
	and	r3, r3, #72
	and	r2, r2, #72
	add	r10, r1, r2
	add	r7, r5, r3
	add	r8, r1, r3
	add	r6, r5, r2
	mov	r3, r10
	mov	r2, r6
	mov	r1, r8
	mov	r0, r7
	bl	xycz_add_c
	mov	r2, r5
	ldr	r3, [sp, #1412]
	ldr	r1, [sp, #48]
	ldr	r0, [sp, #36]
	str	r3, [sp]
	ldr	r3, [sp, #56]
	bl	vli_mod_sub
	mov	r2, #144
	mov	r1, #254
	ldr	r0, [sp, #8]
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2594
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2594:
	ldr	fp, [sp, #36]
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r8
	mov	r1, fp
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, fp
	bl	vli_mmod_fast
	ldr	r3, [sp, #76]
	add	r2, r3, #-1627389952
	ldr	r3, [sp, #72]
	ldrsb	r2, [r2]
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2595
	ldr	r0, [sp, #32]
	bl	__asan_report_load4_noabort
.L2595:
	ldr	r3, [sp, #32]
	mov	r2, #144
	ldr	r0, [sp, #8]
	mov	r1, #254
	ldr	r5, [r3]
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2596
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2596:
	ldr	fp, [sp, #36]
	mov	r2, r5
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r1, fp
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, fp
	bl	vli_mmod_fast
	ldr	r3, [sp, #32]
	add	r0, r3, #8
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2597
	bl	__asan_report_load1_noabort
.L2597:
	ldr	r3, [sp, #32]
	ldr	r1, [sp, #36]
	ldr	r2, [sp, #56]
	ldrb	r3, [r3, #8]	@ zero_extendqisi2
	mov	r0, r1
	bl	vli_mod_inv
	ldr	r1, [sp, #64]
	lsr	r3, r1, #3
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2598
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L2598:
	ldr	r3, [sp, #32]
	mov	r2, #144
	ldr	r0, [sp, #8]
	mov	r1, #254
	ldr	r5, [r3, #4]
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2599
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2599:
	ldr	fp, [sp, #36]
	mov	r2, r5
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r1, fp
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, fp
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2601
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2601:
	ldr	r5, [sp, #36]
	ldr	fp, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r7
	mov	r1, r5
	mov	r0, fp
	bl	vli_mult
	mov	r2, r4
	mov	r1, fp
	mov	r0, r5
	bl	vli_mmod_fast
	mov	r3, r8
	mov	r2, r7
	mov	r1, r10
	mov	r0, r6
	str	r4, [sp]
	bl	xycz_add
	mov	r2, #72
	mov	r1, #254
	mov	r0, r9
	bl	memset
	mov	r2, r4
	mov	r1, r5
	mov	r0, r9
	bl	vli_mod_square_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, fp
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2606
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2606:
	ldr	r5, [sp, #8]
	ldr	r6, [sp, #40]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r9
	mov	r1, r6
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r6
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2607
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2607:
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	ldr	r2, [sp, #36]
	mov	r1, r9
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r9
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r5
	bl	memset
	ldr	r3, [sp, #20]
	ldr	r2, [sp, #16]
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r2, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L2608
	ldr	r0, [sp, #84]
	bl	__asan_report_load1_noabort
.L2608:
	ldr	r6, [sp, #24]
	ldr	r5, [sp, #8]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r9
	mov	r1, r6
	mov	r0, r5
	bl	vli_mult
	mov	r2, r4
	mov	r1, r5
	mov	r0, r6
	bl	vli_mmod_fast
	ldr	r1, [sp, #52]
	lsr	r3, r1, #3
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2609
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L2609:
	ldr	r3, [sp, #52]
	ldr	r1, [sp, #1412]
	mov	r0, #0
	ldr	r5, [r3]
	bl	__sanitizer_cov_trace_const_cmp4
	ldr	r3, [sp, #1412]
	cmp	r3, #0
	bne	.L2610
	b	.L2611
.L2587:
	bl	__sanitizer_cov_trace_pc
	lsr	r5, r8, #6
	ldr	r3, [sp, #28]
	add	r5, r3, r5, lsl #3
	lsr	r3, r5, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2612
	mov	r0, r5
	bl	__asan_report_load8_noabort
.L2612:
	and	r3, r8, #63
	sub	r6, r3, #32
	ldr	r9, [r5]
	rsb	r2, r3, #32
	lsl	r6, r7, r6
	and	r9, r9, r7, lsl r3
	orr	r6, r6, r7, lsr r2
	ldr	r3, [r5, #4]
	mov	r2, r9
	and	r6, r6, r3
	mov	r3, r6
	mov	r0, #0
	mov	r1, #0
	bl	__sanitizer_cov_trace_const_cmp8
	orrs	r3, r9, r6
	mov	r2, r9
	mov	r3, r6
	mov	r0, #0
	mov	r1, #0
	moveq	r5, #1
	movne	r5, #0
	bl	__sanitizer_cov_trace_const_cmp8
	orrs	r9, r9, r6
	movne	r3, #1
	moveq	r3, #0
	sub	r8, r8, #1
	cmp	r3, #0
	movne	r6, #72
	ldr	r3, [sp, #24]
	moveq	r6, #0
	cmp	r5, #0
	movne	r5, #72
	moveq	r5, #0
	add	r10, fp, r6
	add	r9, fp, r5
	add	r6, r3, r6
	add	r5, r3, r5
	mov	r3, r5
	mov	r2, r9
	mov	r1, r6
	mov	r0, r10
	str	r4, [sp]
	bl	xycz_add_c
	mov	r3, r6
	mov	r2, r10
	mov	r1, r5
	mov	r0, r9
	str	r4, [sp]
	bl	xycz_add
	mov	r0, #0
	mov	r1, r8
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r8, #0
	bne	.L2587
	ldr	r9, [sp, #44]
	b	.L2621
.L2611:
	bl	__sanitizer_cov_trace_pc
	mov	r3, #0
	ldr	ip, [sp, #68]
	str	r3, [ip]
	str	r3, [ip, #4]
	str	r3, [ip, #12]
	str	r3, [ip, #16]
	str	r3, [ip, #28]
	str	r3, [ip, #32]
	str	r3, [ip, #40]
	str	r3, [ip, #44]
	str	r3, [ip, #64]
	str	r3, [ip, #68]
	str	r3, [ip, #72]
	str	r3, [ip, #92]
	str	r3, [ip, #96]
	str	r3, [ip, #116]
	str	r3, [ip, #120]
	str	r3, [ip, #124]
	str	r3, [ip, #144]
	str	r3, [ip, #148]
	ldr	r3, .L2798+40
	ldr	r2, [r3]
	ldr	r3, [sp, #1364]
	eors	r2, r3, r2
	mov	r3, #0
	beq	.L2631
	b	.L2797
.L2610:
	ldr	r6, [sp, #40]
	ldr	r7, [sp, #1412]
	mov	r4, #0
.L2624:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2622
	mov	r0, r6
	bl	__asan_report_load8_noabort
.L2622:
	lsr	r3, r5, #3
	add	r3, r3, #-1627389952
	ldrd	r8, [r6], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2623
	mov	r0, r5
	bl	__asan_report_store8_noabort
.L2623:
	add	r4, r4, #1
	mov	r1, r4
	mov	r0, r7
	strd	r8, [r5], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r4, r7
	bne	.L2624
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #52]
	add	r0, r3, #4
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2625
	bl	__asan_report_load4_noabort
.L2625:
	ldr	r3, [sp, #52]
	ldr	r8, [sp, #24]
	ldr	r9, [sp, #1412]
	ldr	r5, [r3, #4]
	mov	r4, #0
.L2628:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r8, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2626
	mov	r0, r8
	bl	__asan_report_load8_noabort
.L2626:
	lsr	r3, r5, #3
	add	r3, r3, #-1627389952
	ldrd	r6, [r8], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L2627
	mov	r0, r5
	bl	__asan_report_store8_noabort
.L2627:
	add	r4, r4, #1
	mov	r1, r4
	mov	r0, r9
	strd	r6, [r5], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r4, r9
	bne	.L2628
	b	.L2611
.L2553:
	bl	__sanitizer_cov_trace_pc
	ldr	r1, [sp, #32]
	lsr	r3, r1, #3
	str	r3, [sp, #76]
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	str	r3, [sp, #72]
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L2629
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L2629:
	ldr	r3, [sp, #32]
	ldr	r1, [sp, #1412]
	mov	r0, #0
	ldr	r5, [r3]
	ldr	r8, .L2798+44
	bl	__sanitizer_cov_trace_const_cmp4
	b	.L2630
.L2797:
	bl	__stack_chk_fail
.L2631:
	add	sp, sp, #1360
	add	sp, sp, #12
	@ sp needed
	mov	r0, #0
	mov	r1, #0
	mov	r2, #0
	mov	r3, #0
	mov	ip, #0
	mov	lr, #0
	pop	{r4, r5, r6, r7, r8, r9, r10, fp, pc}
	.fnend
	.size	ecc_point_mult, .-ecc_point_mult

[-- Attachment #3: ecc_point_mult_limited_inlining.s --]
[-- Type: text/plain, Size: 20637 bytes --]

	.section	.rodata.ecc_point_mult.str1.4,"aMS",%progbits,1
	.align	2
.LC45:
	.ascii	"6 32 72 6 z:1337 144 72 7 t1:1196 256 144 7 rx:1335"
	.ascii	" 464 144 7 ry:1336 672 144 7 sk:1338 880 144 12 pro"
	.ascii	"duct:1028\000"
	.section	.text.ecc_point_mult,"ax",%progbits
	.align	2
	.syntax unified
	.arm
	.type	ecc_point_mult, %function
ecc_point_mult:
.LASANPC2762:
	.fnstart
	@ args = 8, pretend = 0, frame = 1232
	@ frame_needed = 0, uses_anonymous_args = 0
	push	{r4, r5, r6, r7, r8, r9, r10, fp, lr}
	.save {r4, r5, r6, r7, r8, r9, r10, fp, lr}
	.pad #1232
	sub	sp, sp, #1232
	.pad #12
	sub	sp, sp, #12
	push	{lr}
	bl	__gnu_mcount_nc
	str	r3, [sp, #64]
	add	r3, sp, #1232
	add	r3, r3, #8
	bic	lr, r3, #31
	ldr	ip, .L3548
	str	r0, [sp, #44]
	str	r1, [sp, #32]
	str	lr, [sp, #40]
	ldr	r4, [sp, #1280]
	sub	r3, lr, #1120
	str	ip, [lr, #-1120]
	ldr	ip, .L3548+4
	lsr	r3, r3, #3
	str	ip, [lr, #-1116]
	ldr	ip, .L3548+8
	add	r5, r3, #-1627389952
	ldr	r3, .L3548+12
	str	ip, [lr, #-1112]
	mov	r7, lr
	ldr	r0, .L3548+16
	ldr	lr, .L3548+20
	ldr	r1, .L3548+24
	ldr	r6, .L3548+28
	mov	r8, r2
	ldr	r2, .L3548+32
	ldr	fp, [sp, #1284]
	mov	ip, #-234881024
	str	r3, [r5, #28]
	str	r3, [r5, #52]
	str	r3, [r5, #76]
	str	r3, [r5, #80]
	str	r3, [r5, #104]
	ldr	r3, .L3548+36
	str	lr, [r5, #12]
	str	ip, [r5, #24]
	str	r2, [r5, #16]
	str	r0, [r5, #48]
	str	r0, [r5, #100]
	str	r2, [r5, #56]
	str	r2, [r5, #108]
	str	r1, [r5, #128]
	str	r1, [r5, #132]
	str	r6, [r5]
	ldr	r3, [r3]
	str	r3, [sp, #1236]
	mov	r3, #0
	bl	__sanitizer_cov_trace_pc
	sub	r3, r7, #864
	mov	r2, #144
	mov	r1, #254
	mov	r0, r3
	str	r3, [sp, #24]
	bl	memset
	sub	r3, r7, #656
	mov	r2, #144
	mov	r1, #254
	mov	r0, r3
	str	r3, [sp, #36]
	bl	memset
	sub	r3, r7, #1088
	mov	r2, #72
	mov	r1, #254
	mov	r0, r3
	str	r3, [sp, #28]
	bl	memset
	sub	r3, r7, #448
	mov	r2, #144
	mov	r0, r3
	mov	r1, #254
	str	r3, [sp, #48]
	bl	memset
	add	r0, r4, #20
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3325
	bl	__asan_report_load4_noabort
.L3325:
	add	r0, r4, #24
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	ldr	r3, [r4, #20]
	str	r3, [sp, #56]
	beq	.L3326
	bl	__asan_report_load4_noabort
.L3326:
	mov	r1, fp
	mov	r0, #0
	ldr	r7, [r4, #24]
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	fp, #0
	beq	.L3327
	mov	r1, #0
	ldr	r9, [sp, #48]
	mov	r10, r7
	str	r7, [sp, #52]
	mov	r7, r1
	str	r1, [sp, #8]
	str	r9, [sp, #16]
	str	r4, [sp, #60]
	str	fp, [sp, #1284]
	str	r5, [sp, #68]
.L3332:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r8, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3328
	mov	r0, r8
	bl	__asan_report_load8_noabort
.L3328:
	lsr	r3, r10, #3
	add	r3, r3, #-1627389952
	ldm	r8, {r6, fp}
	ldrsb	r3, [r3]
	add	r8, r8, #8
	mov	r5, r10
	cmp	r3, #0
	beq	.L3329
	mov	r0, r10
	bl	__asan_report_load8_noabort
.L3329:
	ldrd	r4, [r5]
	ldr	r3, [sp, #8]
	mov	r0, r6
	adds	r4, r6, r4
	adc	r5, fp, r5
	adds	r4, r4, r3
	adc	r5, r5, #0
	mov	r2, r4
	mov	r3, r5
	mov	r1, fp
	bl	__sanitizer_cov_trace_cmp8
	cmp	fp, r5
	cmpeq	r6, r4
	add	r10, r10, #8
	beq	.L3330
	bl	__sanitizer_cov_trace_pc
	mov	r3, r5
	mov	r1, fp
	mov	r2, r4
	mov	r0, r6
	bl	__sanitizer_cov_trace_cmp8
	cmp	r4, r6
	sbcs	r1, r5, fp
	movcc	r3, #1
	movcs	r3, #0
	str	r3, [sp, #8]
.L3330:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r9, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3331
	mov	r0, r9
	bl	__asan_report_store8_noabort
.L3331:
	add	r7, r7, #1
	strd	r4, [r9]
	ldr	r0, [sp, #1284]
	mov	r1, r7
	bl	__sanitizer_cov_trace_cmp4
	add	r9, r9, #8
	ldr	r3, [sp, #1284]
	cmp	r7, r3
	bne	.L3332
	ldr	r4, [sp, #60]
	ldr	r5, [sp, #68]
	mov	fp, r3
	ldr	r6, [sp, #16]
	ldr	r7, [sp, #52]
	bl	__sanitizer_cov_trace_pc
	mov	r10, #0
	str	r10, [sp, #16]
	str	r4, [sp, #52]
	str	fp, [sp, #1284]
	str	r5, [sp, #60]
	b	.L3341
.L3327:
	bl	__sanitizer_cov_trace_pc
	add	r0, r4, #4
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3334
	bl	__asan_report_load4_noabort
.L3334:
	ldr	r7, [r4, #4]
	ldr	r6, .L3548+40
	mov	r1, r7
	mov	r0, r6
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r7, r6
	beq	.L3335
	ldr	r3, [sp, #40]
	sub	r8, r3, #376
	b	.L3336
.L3412:
	mov	r6, r9
.L3341:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3337
	mov	r0, r6
	bl	__asan_report_load8_noabort
.L3337:
	lsr	r3, r7, #3
	add	r3, r3, #-1627389952
	ldm	r6, {r8, fp}
	ldrsb	r3, [r3]
	add	r9, r6, #8
	mov	r5, r7
	cmp	r3, #0
	beq	.L3338
	mov	r0, r7
	bl	__asan_report_load8_noabort
.L3338:
	ldrd	r4, [r5]
	ldr	r3, [sp, #16]
	mov	r0, r8
	adds	r4, r8, r4
	adc	r5, fp, r5
	adds	r4, r4, r3
	adc	r5, r5, #0
	mov	r2, r4
	mov	r3, r5
	mov	r1, fp
	bl	__sanitizer_cov_trace_cmp8
	cmp	fp, r5
	cmpeq	r8, r4
	add	r7, r7, #8
	beq	.L3339
	bl	__sanitizer_cov_trace_pc
	mov	r3, r5
	mov	r1, fp
	mov	r2, r4
	mov	r0, r8
	bl	__sanitizer_cov_trace_cmp8
	cmp	r4, r8
	sbcs	r1, r5, fp
	movcc	r3, #1
	movcs	r3, #0
	str	r3, [sp, #16]
.L3339:
	add	r6, r6, #72
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3340
	mov	r0, r6
	bl	__asan_report_store8_noabort
.L3340:
	add	r10, r10, #1
	ldr	r0, [sp, #1284]
	mov	r1, r10
	strd	r4, [r9, #64]
	bl	__sanitizer_cov_trace_cmp4
	ldr	r3, [sp, #1284]
	cmp	r10, r3
	bne	.L3412
	ldr	r4, [sp, #52]
	mov	fp, r3
	ldr	r5, [sp, #60]
	bl	__sanitizer_cov_trace_pc
	add	r0, r4, #4
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	ldrsb	r2, [r2]
	ldr	r3, [sp, #8]
	eor	r8, r3, #1
	rsb	r8, r8, #0
	ldr	r3, [sp, #48]
	and	r8, r8, #72
	add	r8, r3, r8
	and	r3, r0, #7
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3343
	bl	__asan_report_load4_noabort
.L3343:
	ldr	r7, [r4, #4]
	ldr	r6, .L3548+40
	mov	r1, r7
	mov	r0, r6
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r7, r6
	beq	.L3344
.L3336:
	bl	__sanitizer_cov_trace_pc
	bl	__sanitizer_cov_trace_pc
	lsl	r7, fp, #6
	sub	r7, r7, #1
	ldr	r1, [sp, #32]
	lsr	r3, r1, #3
	str	r3, [sp, #76]
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	str	r3, [sp, #72]
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3345
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L3345:
	ldr	r3, [sp, #32]
	mov	r1, fp
	mov	r0, #0
	ldr	r6, [r3]
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	fp, #0
	bne	.L3410
	ldr	r3, [sp, #40]
	sub	r2, r3, #792
	str	r2, [sp, #52]
	ldr	r2, [sp, #32]
	sub	r3, r3, #584
	add	r2, r2, #4
	str	r2, [sp, #60]
	str	r3, [sp, #8]
	b	.L3346
.L3410:
	ldr	r3, [sp, #40]
	mov	r10, #0
	sub	r3, r3, #792
	mov	r9, r3
	str	r3, [sp, #52]
.L3349:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3347
	mov	r0, r6
	bl	__asan_report_load8_noabort
.L3347:
	lsr	r3, r9, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	ldrd	r2, [r6], #8
	beq	.L3348
	mov	r0, r9
	strd	r2, [sp, #8]
	bl	__asan_report_store8_noabort
	ldrd	r2, [sp, #8]
.L3348:
	add	r10, r10, #1
	mov	r1, r10
	mov	r0, fp
	strd	r2, [r9], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r10, fp
	bne	.L3349
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #32]
	add	r2, r3, #4
	and	r3, r2, #7
	str	r2, [sp, #60]
	lsr	r2, r2, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3350
	ldr	r0, [sp, #60]
	bl	__asan_report_load4_noabort
.L3350:
	ldr	r3, [sp, #40]
	mov	r10, #0
	sub	r6, r3, #584
	ldr	r3, [sp, #32]
	str	r6, [sp, #8]
	ldr	r9, [r3, #4]
.L3353:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r9, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3351
	mov	r0, r9
	bl	__asan_report_load8_noabort
.L3351:
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	ldrd	r2, [r9], #8
	beq	.L3352
	mov	r0, r6
	strd	r2, [sp, #16]
	bl	__asan_report_store8_noabort
	ldrd	r2, [sp, #16]
.L3352:
	add	r10, r10, #1
	mov	r1, r10
	mov	r0, fp
	strd	r2, [r6], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r10, fp
	bne	.L3353
.L3346:
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #64]
	ldr	r1, [sp, #8]
	ldr	r2, [sp, #24]
	ldr	r0, [sp, #52]
	str	r3, [sp]
	str	r4, [sp, #4]
	ldr	r3, [sp, #36]
	bl	xycz_initial_double
	mov	r1, r7
	mov	r0, #0
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r7, #0
	bgt	.L3354
.L3372:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r8, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3355
	mov	r0, r8
	bl	__asan_report_load8_noabort
.L3355:
	ldr	r3, [r8]
	ldr	ip, [sp, #24]
	and	r3, r3, #1
	eor	r2, r3, #1
	rsb	r2, r2, #0
	rsb	r3, r3, #0
	ldr	lr, [sp, #36]
	and	r3, r3, #72
	and	r2, r2, #72
	add	r0, ip, r3
	add	ip, ip, r2
	add	r1, lr, r3
	str	r4, [sp]
	add	r3, lr, r2
	mov	r2, ip
	str	r0, [sp, #16]
	str	r1, [sp, #48]
	str	ip, [sp, #64]
	str	r3, [sp, #68]
	bl	xycz_add_c
	mov	r1, fp
	mov	r0, #0
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	fp, #0
	bne	.L3360
	b	.L3361
.L3354:
	str	fp, [sp, #1284]
	ldr	fp, [sp, #36]
	mov	r9, #1
	str	r8, [sp, #8]
	str	r5, [sp, #16]
.L3371:
	bl	__sanitizer_cov_trace_pc
	lsr	r5, r7, #6
	ldr	r3, [sp, #8]
	add	r5, r3, r5, lsl #3
	lsr	r3, r5, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3362
	mov	r0, r5
	bl	__asan_report_load8_noabort
.L3362:
	and	r3, r7, #63
	sub	r2, r3, #32
	rsb	r1, r3, #32
	lsl	r2, r9, r2
	ldr	r6, [r5]
	ldr	r8, [r5, #4]
	orr	r2, r2, r9, lsr r1
	and	r6, r6, r9, lsl r3
	and	r8, r8, r2
	mov	r3, r8
	mov	r2, r6
	mov	r0, #0
	mov	r1, #0
	bl	__sanitizer_cov_trace_const_cmp8
	orrs	r3, r6, r8
	mov	r2, r6
	mov	r3, r8
	mov	r0, #0
	mov	r1, #0
	moveq	r5, #1
	movne	r5, #0
	bl	__sanitizer_cov_trace_const_cmp8
	orrs	r8, r6, r8
	movne	r3, #1
	moveq	r3, #0
	sub	r7, r7, #1
	cmp	r3, #0
	movne	r6, #72
	ldr	r3, [sp, #24]
	moveq	r6, #0
	cmp	r5, #0
	movne	r5, #72
	moveq	r5, #0
	add	r10, r3, r6
	add	r8, r3, r5
	add	r6, fp, r6
	add	r5, fp, r5
	mov	r3, r5
	mov	r2, r8
	mov	r1, r6
	mov	r0, r10
	str	r4, [sp]
	bl	xycz_add_c
	mov	r3, r6
	mov	r2, r10
	mov	r1, r5
	mov	r0, r8
	str	r4, [sp]
	bl	xycz_add
	mov	r0, #0
	mov	r1, r7
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	r7, #0
	bne	.L3371
	ldr	r8, [sp, #8]
	ldr	fp, [sp, #1284]
	ldr	r5, [sp, #16]
	b	.L3372
.L3361:
	bl	__sanitizer_cov_trace_pc
	add	r10, r4, #16
	mov	r2, #144
	mov	r1, #254
	lsr	r7, r10, #3
	and	r8, r10, #7
	ldr	r3, [sp, #40]
	sub	r6, r3, #240
	mov	r0, r6
	bl	memset
	add	r2, r7, #-1627389952
	and	r3, r10, #7
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3374
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3374:
	ldr	r9, [sp, #28]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	ldr	r2, [sp, #48]
	mov	r1, r9
	mov	r0, r6
	bl	vli_mult
	mov	r2, r4
	mov	r1, r6
	mov	r0, r9
	bl	vli_mmod_fast
	ldr	r3, [sp, #76]
	add	r2, r3, #-1627389952
	ldr	r3, [sp, #72]
	ldrsb	r2, [r2]
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3375
	ldr	r0, [sp, #32]
	bl	__asan_report_load4_noabort
.L3375:
	ldr	r3, [sp, #32]
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	ldr	r9, [r3]
	bl	memset
	add	r3, r7, #-1627389952
	ldrsb	r3, [r3]
	cmp	r8, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L3376
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3376:
	mov	r2, r9
	ldr	r9, [sp, #28]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r1, r9
	mov	r0, r6
	bl	vli_mult
	mov	r2, r4
	mov	r1, r6
	mov	r0, r9
	bl	vli_mmod_fast
	ldr	r3, [sp, #32]
	add	r0, r3, #8
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3377
	bl	__asan_report_load1_noabort
.L3377:
	ldr	r3, [sp, #32]
	ldr	r1, [sp, #28]
	ldr	r2, [sp, #56]
	ldrb	r3, [r3, #8]	@ zero_extendqisi2
	mov	r0, r1
	bl	vli_mod_inv
	ldr	r1, [sp, #60]
	lsr	r3, r1, #3
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3378
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L3378:
	ldr	r3, [sp, #32]
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	ldr	r9, [r3, #4]
	bl	memset
	add	r3, r7, #-1627389952
	ldrsb	r3, [r3]
	cmp	r8, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L3379
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3379:
	mov	r2, r9
	ldr	r9, [sp, #28]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r1, r9
	mov	r0, r6
	bl	vli_mult
	mov	r2, r4
	mov	r1, r6
	mov	r0, r9
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	bl	memset
	add	r3, r7, #-1627389952
	ldrsb	r3, [r3]
	cmp	r8, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L3381
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3381:
	ldr	r9, [sp, #28]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r1, r9
	ldr	r2, [sp, #16]
	mov	r0, r6
	bl	vli_mult
	mov	r2, r4
	mov	r1, r6
	mov	r0, r9
	bl	vli_mmod_fast
	ldrd	r0, [sp, #64]
	ldr	r3, [sp, #48]
	ldr	r2, [sp, #16]
	str	r4, [sp]
	bl	xycz_add
	mov	r2, #72
	mov	r1, #254
	ldr	r3, [sp, #40]
	sub	r9, r3, #976
	mov	r0, r9
	bl	memset
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	bl	memset
	add	r3, r7, #-1627389952
	ldrsb	r3, [r3]
	cmp	r8, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L3386
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3386:
	ldrb	r2, [r4, #16]	@ zero_extendqisi2
	ldr	r1, [sp, #28]
	mov	r0, r6
	bl	vli_square
	mov	r2, r4
	mov	r1, r6
	mov	r0, r9
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	bl	memset
	add	r3, r7, #-1627389952
	ldrsb	r3, [r3]
	cmp	r8, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L3387
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3387:
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	ldr	r1, [sp, #24]
	mov	r2, r9
	mov	r0, r6
	bl	vli_mult
	mov	r2, r4
	mov	r1, r6
	ldr	r0, [sp, #24]
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	bl	memset
	add	r3, r7, #-1627389952
	ldrsb	r3, [r3]
	cmp	r8, r3
	movlt	r2, #0
	movge	r2, #1
	cmp	r3, #0
	moveq	r2, #0
	cmp	r2, #0
	beq	.L3388
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3388:
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	ldr	r2, [sp, #28]
	mov	r1, r9
	mov	r0, r6
	bl	vli_mult
	add	r7, r7, #-1627389952
	mov	r2, r4
	mov	r1, r6
	mov	r0, r9
	bl	vli_mmod_fast
	mov	r2, #144
	mov	r1, #254
	mov	r0, r6
	bl	memset
	ldrsb	r3, [r7]
	cmp	r8, r3
	movlt	r8, #0
	movge	r8, #1
	cmp	r3, #0
	moveq	r8, #0
	cmp	r8, #0
	beq	.L3389
	mov	r0, r10
	bl	__asan_report_load1_noabort
.L3389:
	ldr	r7, [sp, #36]
	ldrb	r3, [r4, #16]	@ zero_extendqisi2
	mov	r2, r9
	mov	r1, r7
	mov	r0, r6
	bl	vli_mult
	mov	r2, r4
	mov	r1, r6
	mov	r0, r7
	bl	vli_mmod_fast
	ldr	r1, [sp, #44]
	lsr	r3, r1, #3
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3390
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L3390:
	ldr	r3, [sp, #44]
	mov	r1, fp
	mov	r0, #0
	ldr	r6, [r3]
	bl	__sanitizer_cov_trace_const_cmp4
	cmp	fp, #0
	movne	r4, #0
	ldrne	r7, [sp, #24]
	bne	.L3404
	b	.L3392
.L3360:
	ldr	r8, [sp, #28]
	mov	r9, #0
	ldr	r6, [sp, #52]
	str	r9, [sp, #8]
	str	r8, [sp, #52]
	strd	r4, [sp, #80]
.L3397:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3393
	mov	r0, r6
	bl	__asan_report_load8_noabort
.L3393:
	sub	r0, r6, #72
	ldr	r7, [r6]
	lsr	r3, r0, #3
	add	r3, r3, #-1627389952
	ldr	r10, [r6, #4]
	ldrsb	r3, [r3]
	add	r6, r6, #8
	cmp	r3, #0
	beq	.L3394
	bl	__asan_report_load8_noabort
.L3394:
	ldr	r2, [sp, #8]
	ldr	r3, [r6, #-80]
	subs	r4, r7, r2
	ldr	r2, [r6, #-76]
	sbc	r5, r10, #0
	subs	r4, r4, r3
	sbc	r5, r5, r2
	mov	r3, r5
	mov	r2, r4
	mov	r0, r7
	mov	r1, r10
	bl	__sanitizer_cov_trace_cmp8
	cmp	r10, r5
	cmpeq	r7, r4
	beq	.L3395
	bl	__sanitizer_cov_trace_pc
	mov	r3, r5
	mov	r1, r10
	mov	r2, r4
	mov	r0, r7
	bl	__sanitizer_cov_trace_cmp8
	cmp	r7, r4
	sbcs	r1, r10, r5
	movcc	r3, #1
	movcs	r3, #0
	str	r3, [sp, #8]
.L3395:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r8, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3396
	mov	r0, r8
	bl	__asan_report_store8_noabort
.L3396:
	add	r9, r9, #1
	strd	r4, [r8]
	mov	r1, r9
	mov	r0, fp
	bl	__sanitizer_cov_trace_cmp4
	cmp	r9, fp
	add	r8, r8, #8
	bne	.L3397
	ldr	r7, [sp, #52]
	ldrd	r4, [sp, #80]
	bl	__sanitizer_cov_trace_pc
	mov	r3, #0
	mov	r0, #0
	mov	r1, #0
	ldr	r6, [sp, #8]
	mov	r2, r6
	bl	__sanitizer_cov_trace_const_cmp8
	orrs	r9, r6, #0
	beq	.L3361
	ldr	r9, [sp, #56]
	mov	r10, #0
	str	r10, [sp, #8]
	str	r4, [sp, #52]
	str	r5, [sp, #80]
.L3401:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r7, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3398
	mov	r0, r7
	bl	__asan_report_load8_noabort
.L3398:
	lsr	r3, r9, #3
	add	r3, r3, #-1627389952
	ldm	r7, {r6, r8}
	ldrsb	r3, [r3]
	add	r7, r7, #8
	mov	r5, r9
	cmp	r3, #0
	beq	.L3399
	mov	r0, r9
	bl	__asan_report_load8_noabort
.L3399:
	ldrd	r4, [r5]
	ldr	r3, [sp, #8]
	mov	r0, r6
	adds	r4, r6, r4
	adc	r5, r8, r5
	adds	r4, r4, r3
	adc	r5, r5, #0
	mov	r2, r4
	mov	r3, r5
	mov	r1, r8
	bl	__sanitizer_cov_trace_cmp8
	cmp	r8, r5
	cmpeq	r6, r4
	add	r9, r9, #8
	beq	.L3400
	bl	__sanitizer_cov_trace_pc
	mov	r3, r5
	mov	r1, r8
	mov	r2, r4
	mov	r0, r6
	bl	__sanitizer_cov_trace_cmp8
	cmp	r4, r6
	sbcs	r1, r5, r8
	movcc	r3, #1
	movcs	r3, #0
	str	r3, [sp, #8]
.L3400:
	add	r10, r10, #1
	bl	__sanitizer_cov_trace_pc
	mov	r1, r10
	mov	r0, fp
	strd	r4, [r7, #-8]
	bl	__sanitizer_cov_trace_cmp4
	cmp	r10, fp
	bne	.L3401
	ldr	r4, [sp, #52]
	ldr	r5, [sp, #80]
	b	.L3361
.L3392:
	bl	__sanitizer_cov_trace_pc
	mov	r2, r5
	mov	r3, #0
	str	r3, [r2], #12
	str	r3, [r5, #12]
	str	r3, [r2, #4]
	str	r3, [r5, #24]
	str	r3, [r5, #28]
	str	r3, [r5, #48]
	str	r3, [r5, #52]
	str	r3, [r5, #56]
	str	r3, [r5, #76]
	str	r3, [r5, #80]
	str	r3, [r5, #100]
	str	r3, [r5, #104]
	str	r3, [r5, #108]
	str	r3, [r5, #128]
	str	r3, [r5, #132]
	ldr	r3, .L3548+36
	ldr	r2, [r3]
	ldr	r3, [sp, #1236]
	eors	r2, r3, r2
	mov	r3, #0
	beq	.L3411
	b	.L3547
.L3549:
	.align	2
.L3548:
	.word	1102416563
	.word	.LC45
	.word	.LASANPC2762
	.word	-218959118
	.word	-219021312
	.word	-218959360
	.word	-202116109
	.word	-235802127
	.word	62194
	.word	__stack_chk_guard
	.word	521
.L3404:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r7, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3402
	mov	r0, r7
	bl	__asan_report_load8_noabort
.L3402:
	lsr	r3, r6, #3
	add	r3, r3, #-1627389952
	ldrd	r8, [r7], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3403
	mov	r0, r6
	bl	__asan_report_store8_noabort
.L3403:
	add	r4, r4, #1
	mov	r1, r4
	mov	r0, fp
	strd	r8, [r6], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r4, fp
	bne	.L3404
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #44]
	add	r0, r3, #4
	and	r3, r0, #7
	lsr	r2, r0, #3
	add	r2, r2, #-1627389952
	add	r3, r3, #3
	ldrsb	r2, [r2]
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3405
	bl	__asan_report_load4_noabort
.L3405:
	ldr	r3, [sp, #44]
	ldr	r9, [sp, #36]
	mov	r4, #0
	ldr	r8, [r3, #4]
.L3408:
	bl	__sanitizer_cov_trace_pc
	lsr	r3, r9, #3
	add	r3, r3, #-1627389952
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3406
	mov	r0, r9
	bl	__asan_report_load8_noabort
.L3406:
	lsr	r3, r8, #3
	add	r3, r3, #-1627389952
	ldrd	r6, [r9], #8
	ldrsb	r3, [r3]
	cmp	r3, #0
	beq	.L3407
	mov	r0, r8
	bl	__asan_report_store8_noabort
.L3407:
	add	r4, r4, #1
	mov	r1, r4
	mov	r0, fp
	strd	r6, [r8], #8
	bl	__sanitizer_cov_trace_cmp4
	cmp	r4, fp
	bne	.L3408
	b	.L3392
.L3335:
	bl	__sanitizer_cov_trace_pc
	ldr	r3, [sp, #64]
	ldr	r6, [sp, #40]
	ldr	r2, [sp, #24]
	sub	r0, r6, #792
	sub	r1, r6, #584
	stm	sp, {r3, r4}
	ldr	r3, [sp, #36]
	str	r0, [sp, #52]
	bl	xycz_initial_double
	mov	r1, r7
	mov	r0, #0
	bl	__sanitizer_cov_trace_const_cmp4
	sub	r8, r6, #376
	ldr	r3, [sp, #32]
	lsr	r2, r3, #3
	str	r2, [sp, #76]
	and	r2, r3, #7
	add	r3, r3, #4
	str	r2, [sp, #72]
	str	r3, [sp, #60]
	b	.L3354
.L3344:
	bl	__sanitizer_cov_trace_pc
	ldr	r1, [sp, #32]
	lsr	r3, r1, #3
	str	r3, [sp, #76]
	add	r3, r3, #-1627389952
	ldrsb	r2, [r3]
	and	r3, r1, #7
	str	r3, [sp, #72]
	add	r3, r3, #3
	cmp	r3, r2
	movlt	r3, #0
	movge	r3, #1
	cmp	r2, #0
	moveq	r3, #0
	cmp	r3, #0
	beq	.L3409
	mov	r0, r1
	bl	__asan_report_load4_noabort
.L3409:
	ldr	r3, [sp, #32]
	mov	r1, fp
	mov	r0, #0
	ldr	r6, [r3]
	ldr	r7, .L3548+40
	bl	__sanitizer_cov_trace_const_cmp4
	b	.L3410
.L3547:
	bl	__stack_chk_fail
.L3411:
	add	sp, sp, #1232
	add	sp, sp, #12
	@ sp needed
	mov	r0, #0
	mov	r1, #0
	mov	r2, #0
	mov	r3, #0
	mov	ip, #0
	mov	lr, #0
	pop	{r4, r5, r6, r7, r8, r9, r10, fp, pc}
	.fnend
	.size	ecc_point_mult, .-ecc_point_mult

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Arnd Bergmann]

On Mon, Apr 13, 2026, at 21:46, Lukas Wunner wrote:
> On Mon, Apr 13, 2026 at 05:42:39PM +0200, Arnd Bergmann wrote:
>> On Wed, Apr 8, 2026, at 15:36, Lukas Wunner wrote:
>
> Attached please find the Assembler output created by gcc -save-temps,
> both the original version and the one with limited inlining.
>
> The former requires a 1360 bytes stack frame, the latter 1232 bytes.
> E.g. xycz_initial_double() is not inlined into ecc_point_mult(),
> together with all its recursive baggage, so the latter version
> contains two branch instructions to that function which the former
> (original) version does not contain.

Thanks!

So it indeed appears that the problem does not go away but only
stays below the arbitrary threshold of 1280 bytes (which was
recently raised). I would not trust that to actually be the
case across all architectures then, as there are some targets
like mips or parisc tend to use even more stack space than
arm. With your current patch, that means there is a good chance
the problem will come back later.

> At the beginning of the function, it looks like the same register values
> are stored to multiple locations on the stack.  I assume that's what you
> mean by awful code generation?  This odd behavior seems more subdued in
> the version with limited inlining.

Right. As far as I can tell, the source code is heavily optimized
for performance, but with the sanitizer active this would likely
be several times slower, both from the actual sanitizing and
from the register spilling. I can see how the use of 'u64'
arrays makes this harder for a 32-bit target with limited
available registers.

     Arnd

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[Lukas Wunner]

On Mon, Apr 13, 2026 at 10:32:24PM +0200, Arnd Bergmann wrote:
> On Mon, Apr 13, 2026, at 21:46, Lukas Wunner wrote:
> > On Mon, Apr 13, 2026 at 05:42:39PM +0200, Arnd Bergmann wrote:
> > > On Wed, Apr 8, 2026, at 15:36, Lukas Wunner wrote:
> > Attached please find the Assembler output created by gcc -save-temps,
> > both the original version and the one with limited inlining.
> >
> > The former requires a 1360 bytes stack frame, the latter 1232 bytes.
> > E.g. xycz_initial_double() is not inlined into ecc_point_mult(),
> > together with all its recursive baggage, so the latter version
> > contains two branch instructions to that function which the former
> > (original) version does not contain.
> 
> So it indeed appears that the problem does not go away but only
> stays below the arbitrary threshold of 1280 bytes (which was
> recently raised). I would not trust that to actually be the
> case across all architectures then, as there are some targets
> like mips or parisc tend to use even more stack space than
> arm. With your current patch, that means there is a good chance
> the problem will come back later.

The only 32-bit architectures with HAVE_ARCH_KASAN are:
arm powerpc xtensa

I've cross-compiled ecc.o successfully in an allmodconfig build for
powerpc and xtensa, so arm seems to be the only architecture affected
by the large stack frame issue.

Maybe mips and parisc will see the issue as well but they'd have to
support KASAN first.

The problem is that gcc *knows* that it should warn when the stack
goes above CONFIG_FRAME_WARN and that warning is even promoted to
an error, but gcc happily keeps inlining stuff and goes beyond that
limit.  My expectation is it should stop inlining before that happens.
clang doesn't have the same problem.

Completely disabling KASAN for this file doesn't seem like a good option
as this is security-relevant code.  On the other hand disabling inlining
for this file isn't great either because I recall Google is dogfooding
KASAN on internally used phones, I imagine it would ruin performance
for such use cases (granted those are likely arm64 devices).

*Limiting* inlining strikes a middle ground between those two extremes.

And I don't want to annotate individual functions as noinline only
because gcc does stupid things on a single architecture.

Thanks,

Lukas

Re: [PATCH] crypto: ecc - Unbreak the build on arm with CONFIG_KASAN_STACK=y

[David Laight]

On Mon, 13 Apr 2026 22:32:24 +0200
"Arnd Bergmann" <arnd@arndb.de> wrote:

> On Mon, Apr 13, 2026, at 21:46, Lukas Wunner wrote:
> > On Mon, Apr 13, 2026 at 05:42:39PM +0200, Arnd Bergmann wrote:  
> >> On Wed, Apr 8, 2026, at 15:36, Lukas Wunner wrote:  
> >
> > Attached please find the Assembler output created by gcc -save-temps,
> > both the original version and the one with limited inlining.
> >
> > The former requires a 1360 bytes stack frame, the latter 1232 bytes.
> > E.g. xycz_initial_double() is not inlined into ecc_point_mult(),
> > together with all its recursive baggage, so the latter version
> > contains two branch instructions to that function which the former
> > (original) version does not contain.  
> 
> Thanks!
> 
> So it indeed appears that the problem does not go away but only
> stays below the arbitrary threshold of 1280 bytes (which was
> recently raised). I would not trust that to actually be the
> case across all architectures then, as there are some targets
> like mips or parisc tend to use even more stack space than
> arm. With your current patch, that means there is a good chance
> the problem will come back later.

Not only that, the 'stack frome size' is just a proxy for total
stack use - which is a lot harder to calculate.
I've a cunning plan to use clangs function prototype hashing
to do a static stack calculation that includes indirect calls.
(I did one many years ago for some embedded code that had none.)
I suspect it will find all sorts of code paths that 'blow' the
kernel stack out of the water.
A good bet will be snprintf() calls in unusual error paths
(even after ignoring recursive snprintf() calls and all the %px
modifiers).

> > At the beginning of the function, it looks like the same register values
> > are stored to multiple locations on the stack.  I assume that's what you
> > mean by awful code generation?  This odd behavior seems more subdued in
> > the version with limited inlining.  
> 
> Right. As far as I can tell, the source code is heavily optimized
> for performance, but with the sanitizer active this would likely
> be several times slower, both from the actual sanitizing and
> from the register spilling. I can see how the use of 'u64'
> arrays makes this harder for a 32-bit target with limited
> available registers.

gcc make a right 'pigs breakfast' of handling u64 items on 32bit.
It gets really horrid on x86 (which has 8 registers including %sp
and %bp).
I got the impression it sometimes treats a u64 as being two 32bit
values, and other times as a 64bit value held in two registers.
The former tends to generate better code, but that latter happens
if an asm() block (or probably anything else) ends up with an 'A'
constraint for a value in %edx:%eax.
It will spill constant zero words to stack, and do multiplies by
values that are constant zero.
(I think the code generated for a single call to mul_64_64()
will show it all.)

I've just looked at that source.
It seems to be doing 'very wide' arithmetic using u64[].
That will be really horrid on 32bit - it needs to use u32[].

Stopping some of those function being inlined will help.
Even on 64bit I doubt it'll make that much difference to
overall performance.

	David

> 
>      Arnd
>