wolfssl: use -fomit-frame-pointer to fix asm error
authorEneas U de Queiroz <cotequeiroz@gmail.com>
Tue, 26 May 2020 13:45:22 +0000 (10:45 -0300)
committerPetr Štetiar <ynezz@true.cz>
Wed, 3 Jun 2020 14:49:28 +0000 (16:49 +0200)
32-bit x86 fail to compile fast-math feature when compiled with frame
pointer, which uses a register used in a couple of inline asm functions.

Previous versions of wolfssl had this by default.  Keeping an extra
register available may increase performance, so it's being restored for
all architectures.

Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
package/libs/wolfssl/Makefile

index b186a087e70c6c870e2c8ac6ad56d0a608ae2797..159cfbc53f74a481ed5fa037d82892ede138a900 100644 (file)
@@ -9,7 +9,7 @@ include $(TOPDIR)/rules.mk
 
 PKG_NAME:=wolfssl
 PKG_VERSION:=4.4.0-stable
-PKG_RELEASE:=1
+PKG_RELEASE:=2
 
 PKG_SOURCE:=$(PKG_NAME)-$(PKG_VERSION).tar.gz
 PKG_SOURCE_URL:=https://github.com/wolfSSL/wolfssl/archive/v$(PKG_VERSION)
@@ -56,7 +56,7 @@ define Package/libwolfssl/config
        source "$(SOURCE)/Config.in"
 endef
 
-TARGET_CFLAGS += $(FPIC) -DFP_MAX_BITS=8192
+TARGET_CFLAGS += $(FPIC) -DFP_MAX_BITS=8192 -fomit-frame-pointer
 
 # --enable-stunnel needed for OpenSSL API compatibility bits
 CONFIGURE_ARGS += \