Re: [cryptopp-users] Add -fno-lto for GCC and compatibles

2021-05-05 Thread Uri Blumenthal
I don't see a good alternative. -fno-lto seems it. 

> On May 5, 2021, at 14:23, Jeffrey Walton  wrote:
> 
> ´╗┐Hi Everyone,
> 
> LTO is slowly accumulating bugs. From what I have seen in the
> post-mortems, it looks like LTO is either (1) results in a compile
> error (older compilers) or (2) results in a runtime error (newer
> compilers).
> 
> For (2), it appears LTO mishandles code with GCC extended inline
> assembly. To build the library with LTO, you have to use
> -DCRYPTOPP_DISABLE_ASM. That gets us past the self tests but causes a
> slowdown in the library.
> 
> Worse, most users and distros are not aware of the LTO problems so the
> library gets built with it but without -DCRYPTOPP_DISABLE_ASM.
> 
> I think we should add -fno-lto when GCC and compatibles support it.
> 
> What do you think?
> 
> Jeff
> 
> -- 
> You received this message because you are subscribed to the Google Groups 
> "Crypto++ Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to cryptopp-users+unsubscr...@googlegroups.com.
> To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cryptopp-users/CAH8yC8nJERS3-Y4KNZfz15fu9PJDJpE%2BZM-KmV7RcMv-oTth8w%40mail.gmail.com.

-- 
You received this message because you are subscribed to the Google Groups 
"Crypto++ Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to cryptopp-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/cryptopp-users/C7B619BE-EC36-4399-A44F-98D1302B56F3%40gmail.com.


smime.p7s
Description: S/MIME cryptographic signature


[cryptopp-users] Add -fno-lto for GCC and compatibles

2021-05-05 Thread Jeffrey Walton
Hi Everyone,

LTO is slowly accumulating bugs. From what I have seen in the
post-mortems, it looks like LTO is either (1) results in a compile
error (older compilers) or (2) results in a runtime error (newer
compilers).

For (2), it appears LTO mishandles code with GCC extended inline
assembly. To build the library with LTO, you have to use
-DCRYPTOPP_DISABLE_ASM. That gets us past the self tests but causes a
slowdown in the library.

Worse, most users and distros are not aware of the LTO problems so the
library gets built with it but without -DCRYPTOPP_DISABLE_ASM.

I think we should add -fno-lto when GCC and compatibles support it.

What do you think?

Jeff

-- 
You received this message because you are subscribed to the Google Groups 
"Crypto++ Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to cryptopp-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/cryptopp-users/CAH8yC8nJERS3-Y4KNZfz15fu9PJDJpE%2BZM-KmV7RcMv-oTth8w%40mail.gmail.com.