<div dir="ltr"><div dir="ltr"><div dir="auto">I moved the implementation from common.c to common.h and marked them static inline.</div><div dir="auto"><br></div><div>This seems to be the most portable approach for inlining (<a href="https://www.greenend.org.uk/rjk/tech/inline.html">https://www.greenend.org.uk/rjk/tech/inline.html</a>).</div><div><br></div><div>M.<br></div></div></div><br><div class="gmail_quote"><div dir="ltr">On Mon, 1 Oct 2018, 17:26 Brendan Dolan-Gavitt, <<a href="mailto:brendandg@nyu.edu" target="_blank">brendandg@nyu.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">What's involved in inlining them? Simply marking them as "inline" in the header?<br>
<br>
On Mon, Oct 1, 2018 at 9:39 AM, Manolis Stamatogiannakis<br>
<<a href="mailto:mstamat@gmail.com" rel="noreferrer" target="_blank">mstamat@gmail.com</a>> wrote:<br>
> I had this in the back of my mind for some time, but today I gave it a try<br>
> to see the effects.<br>
><br>
> Many functions declared in panda/common.h are only thin wrappers around<br>
> other functions. E.g.:<br>
> int panda_physical_memory_rw(hwaddr addr, uint8_t *buf, int len, int<br>
> is_write);<br>
> hwaddr panda_virt_to_phys(CPUState *env, target_ulong addr);<br>
> int panda_virtual_memory_rw(CPUState *env, target_ulong addr, uint8_t *buf,<br>
> int len, int is_write);<br>
> int panda_virtual_memory_read(CPUState *env, target_ulong addr, uint8_t<br>
> *buf, int len);<br>
> int panda_virtual_memory_write(CPUState *env, target_ulong addr, uint8_t<br>
> *buf, int len);<br>
><br>
> Is there any reason these have not been implemented as inlines?<br>
><br>
> After inlining them (a 5' task), my reply replay ran ~5% faster (141.7sec -><br>
> 133.7sec). Not a huge improvement, but given the effort required for the<br>
> change, one can't complain.<br>
><br>
> Thanks,<br>
> Manolis<br>
><br>
> _______________________________________________<br>
> panda-users mailing list<br>
> <a href="mailto:panda-users@mit.edu" rel="noreferrer" target="_blank">panda-users@mit.edu</a><br>
> <a href="https://urldefense.proofpoint.com/v2/url?u=http-3A__mailman.mit.edu_mailman_listinfo_panda-2Dusers&d=DwICAg&c=slrrB7dE8n7gBJbeO0g-IQ&r=A4wu5Zmpus3hDmokNWeJTO0SLjrxguzCAxn30Hc-o48&m=HSkp0HaNEn4JxGH47OBshPtVt8yawjH9TlZ7cW5xaH4&s=Z6_aPLmeXQf7Lrg7FEBZtXPy2bSMUjaDmTTrkGjskP4&e=" rel="noreferrer noreferrer" target="_blank">https://urldefense.proofpoint.com/v2/url?u=http-3A__mailman.mit.edu_mailman_listinfo_panda-2Dusers&d=DwICAg&c=slrrB7dE8n7gBJbeO0g-IQ&r=A4wu5Zmpus3hDmokNWeJTO0SLjrxguzCAxn30Hc-o48&m=HSkp0HaNEn4JxGH47OBshPtVt8yawjH9TlZ7cW5xaH4&s=Z6_aPLmeXQf7Lrg7FEBZtXPy2bSMUjaDmTTrkGjskP4&e=</a><br>
><br>
<br>
<br>
<br>
-- <br>
Brendan Dolan-Gavitt<br>
Assistant Professor, Department of Computer Science and Engineering<br>
NYU Tandon School of Engineering<br>
</blockquote></div>