On 3/30/23 18:48, Viktor Dukhovni wrote:
On Thu, Mar 30, 2023 at 05:50:50PM -0400, Dennis Clarke via openssl-users wrote:
This is exactly the sort of thing this email list for ordinary users
was intended for. Custom designed for this sort of off the wall crazy
question. A github bug report would be a vulgar display of stupid.
Your question is a bit surprising, because those enterprising enough to
operate what is now exotic hardware, are usually the sort of users who
answer rather than ask this sort of question...
That is true and perhaps I was just lazy. HOWEVER this does present a
fine demonstration of just how valuable this email list is to ordinary
users. Almost anything can happen and reasonable questions, even from
the exotic world of hardware no one else has ( or remembers! ) may arrive.
That said, the header in question occurs in exactly one place in
OpenSSL:
Sadly, yes, I did the grand "find . -type f | xargs grep foo" right
*after* sending this email.
Yikes.
Tazer me gently.
crypto/threads_pthread.c-#if defined(__sun)
crypto/threads_pthread.c:# include <atomic.h>
.
.
.
crypto/threads_pthread.c-#endif
To be used on some unspecified "__sun" systems with some unspecified
toolchain. You'll discover what's missing without it by removing the
include, and seeing what breaks, and then fix that, or perhaps switching
toolchain versions to one that provides that header.
Another thing to do could be to move the include below (inside):
#if defined(OPENSSL_THREADS) && !defined(CRYPTO_TDEBUG) && !defined(OPENSSL_SYS_WINDOWS)
and Configure "no-threads", assuming you don't have to have threads.
It may take a week to compile but I shall report back that OpenSSL 3.1.0
can be built and tested on a toaster. Also a thirty year old Sun 32-bit
SparcStation 20. Which actually is a nice foot warmer also.
--
Dennis Clarke
RISC-V/SPARC/PPC/ARM/CISC
Four decades in production systems.
