Building OpenSSL 1.1.1h with gcc9 on linux creates build error in b_addr.c · Issue #13049

link管理

链接快照平台

输入网页链接，自动生成快照
标签化管理网页链接

相关文章推荐

爱听歌的抽屉 · allure测试报告的友好展示方式开发者社区· 1 月前 ·

眉毛粗的豆浆 · 解决Jenkins中Node和Yarn路径问 ...· 4 周前 ·

悲伤的小刀 · 猎豹免费WiFi5.1.9257.1447猎 ...· 3 月前 ·

沉着的汉堡包 · 法国石楠根烟斗品牌 - ...· 3 月前 ·

性感的大象 · 如何将字符串数组转换为对象？_如何将数组([ ...· 5 月前 ·

面冷心慈的大海 · [NET-408] problem ...· 5 月前 ·

率性的大熊猫 · Constant Field Values ...· 7 月前 ·

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement . We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

We recently performed our own OpenSSL build based on the 1.1.1h sources from this repo as static library and ensured that we have the same compiler settings used for the build as we use elsewhere for our C++ based code. This means that we also activated _POSIX_C_SOURCE=200809L while doing so. With these settings everything builds fine except the file crypto/bio/b_addr.c, which gives us the following diagnostics for gcc9:

[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c: In function ‘addr_strings’:
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:198:19: error: ‘NI_MAXHOST’ undeclared (first use in this function)
[INFO]   198 |         char host[NI_MAXHOST] = "", serv[NI_MAXSERV] = "";
[INFO]       |                   ^~~~~~~~~~
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:198:19: note: each undeclared identifier is reported only once for each function it appears in
[INFO] [ 15%] Building C object crypto/CMakeFiles/crypto.dir/bio/bf_null.c.o
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:198:42: error: ‘NI_MAXSERV’ undeclared (first use in this function)
[INFO]   198 |         char host[NI_MAXHOST] = "", serv[NI_MAXSERV] = "";
[INFO]       |                                          ^~~~~~~~~~
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:198:37: warning: unused variable ‘serv’ [-Wunused-variable]
[INFO]   198 |         char host[NI_MAXHOST] = "", serv[NI_MAXSERV] = "";
[INFO]       |                                     ^~~~
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:198:14: warning: unused variable ‘host’ [-Wunused-variable]
[INFO]   198 |         char host[NI_MAXHOST] = "", serv[NI_MAXSERV] = "";
[INFO]       |              ^~~~
[INFO] [ 15%] Building C object crypto/CMakeFiles/crypto.dir/bio/bio_cb.c.o
[INFO] In file included from /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/include/internal/cryptlib.h:25,
[INFO]                  from /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/bio_local.h:89,
[INFO]                  from /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:13:
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c: In function ‘BIO_lookup_ex’:
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:811:52: error: ‘h_errno’ undeclared (first use in this function); did you mean ‘errno’?
[INFO]   811 |                 SYSerr(SYS_F_GETHOSTBYNAME, 1000 + h_errno);
[INFO]       |                                                    ^~~~~~~
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/include/openssl/err.h:29:61: note: in definition of macro ‘ERR_PUT_error’
[INFO]    29 | #  define ERR_PUT_error(a,b,c,d,e)        ERR_put_error(a,b,c,d,e)
[INFO]       |                                                             ^
[INFO] /home/jenkins/jenkins/workspace/dmb1/com.github.openssl-linux-x86_64-gcc9/src/crypto/bio/b_addr.c:811:17: note: in expansion of macro ‘SYSerr’
[INFO]   811 |                 SYSerr(SYS_F_GETHOSTBYNAME, 1000 + h_errno);
[INFO]       |                 ^~~~~~
[INFO] make[2]: *** [crypto/CMakeFiles/crypto.dir/build.make:1419: crypto/CMakeFiles/crypto.dir/bio/b_addr.c.o] Error 1
The errors all can be attributed to the same cause:
Regarding NI_MAXHOST: According to the man pages:
Since glibc 2.8, these definitions are exposed only if suitable feature test macros are defined, namely: _GNU_SOURCE,

_DEFAULT_SOURCE (since glibc 2.19), or (in glibc versions up to and including 2.19) _BSD_SOURCE or _SVID_SOURCE.
Regarding  h_errno: There seem to exist similar need to add _GNU_SOURCE
We could resolve all compile problems by applying the following changes:
--- src/crypto/bio/b_addr.c	
+++ src/crypto/bio/b_addr.c	
@@ -7,6 +7,10 @@
  * https://www.openssl.org/source/license.html
+#ifndef _GNU_SOURCE
+# define _GNU_SOURCE
+#endif
 #include <assert.h>
 #include <string.h>
This fix seems to be in sync with some other translation units of OpenSSL which also conditionally define _GNU_SOURCE at the very beginning.
          Hmmm, if those macros aren't guaranteed to be available everywhere, wouldn't it make more sense to add this directly after the includes?
#ifndef NI_MAXHOST
# define NI_MAXHOST 1025
#endif
#ifndef NI_MAXSERV
# define NI_MAXSERV 32
#endif
The users can still choose to use _GNU_SOURCE when configuring:
./Configure CFLAGS='-D_GNU_SOURCE'
          First: That wouldn't solve the h_errno problem. Second: I don't think that it is a good recommendation to say that user code should add -D_GNU_SOURCE to the compile flags, because this is a define that is usually an implementation detail and should not be set (or unset) universally. It is typically relevant for a specific translation unit and this unit should control this flag.

Last but not least is the suggested fix consistent with the approaches used in other translation units of OpenSSL which do the same thing.
We could resolve all compile problems by applying the following changes:
Can you supply a pull request. Presumably we would need to make the same change against master.
I made a pull request with "CLA: trivial" as a separate line, but that was rejected. Is this really considered more than a "one-liner" pull-request?
We could resolve all compile problems by applying the following changes:
Can you supply a pull request. Presumably we would need to make the same change against master.
I made a pull request with "CLA: trivial" as a separate line, but that was rejected. Is this really considered more than a "one-liner" pull-request?
The automation looks at the actual git commit message, not the pull request summary
I made a pull request with "CLA: trivial" as a separate line, but that was rejected. Is this really considered more than a "one-liner" pull-request?
The automation looks at the actual git commit message, not the pull request summary
Thanks! I made now a new pull request (due to problems with amending the original commit) and I ensured that both
CLA: trivial
Fixes #13049
appear as separate lines, but I still get the same error. Apologies if I'm overlooking something obvious. What did I do wrong this time? Thanks for your help.
      Since glibc 2.8, these defines like `NI_MAXHOST` are exposed only
if suitable feature test macros are defined, namely: _GNU_SOURCE,
_DEFAULT_SOURCE (since glibc 2.19), or _BSD_SOURCE or _SVID_SOURCE
(before glibc 2.19), see GETNAMEINFO(3).
CLA: trivial
Fixes openssl#13049
      Since glibc 2.8, these defines like `NI_MAXHOST` are exposed only
if suitable feature test macros are defined, namely: _GNU_SOURCE,
_DEFAULT_SOURCE (since glibc 2.19), or _BSD_SOURCE or _SVID_SOURCE
(before glibc 2.19), see GETNAMEINFO(3).
CLA: trivial
Fixes #13049
Reviewed-by: Richard Levitte <[email protected]>
Reviewed-by: Matt Caswell <[email protected]>
Reviewed-by: Matthias St. Pierre <[email protected]>
(Merged from #13054)
(cherry picked from commit 99501e8)