From mboxrd@z Thu Jan 1 00:00:00 1970 From: Cyril Hrubis Date: Wed, 20 Feb 2019 16:12:50 +0100 Subject: [LTP] [PATCH v2] syscalls: add syscall syncfs test In-Reply-To: <20190220131443.GA5256@rei> References: <1550215053-6795-1-git-send-email-sumit.garg@linaro.org> <5C667060.7020405@cn.fujitsu.com> <20190215121611.GA14270@rei> <20190215132251.GB26339@rei> <4cae2a11-47c2-c319-19cf-9a8fecc7793d@google.com> <20190218115746.GA25025@rei.lan> <6c93cc5b-f871-0160-0e81-0d964059cdf2@google.com> <20190220131443.GA5256@rei> Message-ID: <20190220151248.GB32140@rei> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ltp@lists.linux.it Hi! > > Sure. My concern is being able to test syscalls in Android where the C > > library may not have some wrappers. So far these have all been cases > > where one can just replace the library call with the direct syscall, as > > a number of recent patches have done. If I run into a case where more > > substantial library support is needed maybe we'll just have to focus on > > getting that into bionic or look at other options. > > I guess that we can also auto-generate fallback syscall wrappers for > these cases so that we don't have to bother dealing with this in the > actual testcases. Should be as easy as listing the function prototypes > in a file and a few lines of shell. Thinking of this, generating fallbacks is the easy part, figuring out when to use them is the complex one. I wanted to use a trick with weak function symbols so that the linker would pick up these fallbacks only if there was no system defintion but that unfortunately does not work because glibc uses weak symbols for symbol versioning. I guess that we can use autoconf to generate fallback functions automatically if syscall wrapper is found to be missing, but that would be a bit ugly code. Also do you even use the configure script on android builds? -- Cyril Hrubis chrubis@suse.cz