Re: [PATCH 1/2] Don't declare xcb_req as static
Keith Packard <[email protected]>
| Newsgroups | gmane.comp.freedesktop.xcb |
|---|---|
| Message-ID | <[email protected]> |
Eric Anholt <[email protected]> writes: > Increasing the executed-code size by 4% here makes me nervous Overall memory usage in the system will be less as you get rid of the data segment duplication in every process. > Is the problem just the relocations in that data? Yes, even if the xcb_extension_t is also const, you've got a relocation to deal with. > Would pulling the req->ext out of xcb_protocol_request and passing it > as another parameter of a new xcb_send_request variant also fix it, > while letting us keep the remainder of the xcb_protocol_request_t > static? Yes, it looks like that should work. I note that there's also a global mutex lock for every extension request which would be nice to eliminate at some point :-) -- -keith _______________________________________________ Xcb mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/xcb
signature.asc
(application/pgp-signature, 832 B)
-----BEGIN PGP SIGNATURE----- iQIzBAEBCAAdFiEEw4O3eCVWE9/bQJ2R2yIaaQAAABEFAllT5zoACgkQ2yIaaQAA ABFLXA//aUTnxPKB7WtX41zm4fLzBCMTLitv3D/Z1ihmxfyp1zrqhdqg+fr0DcoP 0hdEjrvD9uo21ZJ7weeFvsVONaaatPu0dGpHnnfXvhxjkUVFPFujjx+B3eaTA0i5 okVXcyFWs0lbiVsvU8qiEbiZOMff8X4Wyptmja2SpsGpZ9bCW6lMt21TW4jfgM5z Ima7/7+39B3Tf6WbKT9BiZ3rpkSF5Pnd6AtGBlgYp+Bh0AN5At03dPRcKRnBAxJQ elQOPa9pdAeQXCzgJbVKmQwrhHJbRnin5BgEteIkMjEcLs8jwwuyH/R07U4/UjYi mj7z+mLklc50u3YKGfknwhREFci4OXvha22u5VAgcFiFDr/7y+GFSgrdnHurSXxG F3xtdAyWEOZcQoTtS216xbR7RbGGpIqcl6ElsON1n69CqQyq8OCczJJD7pIHLjwG h/TjNhMbXzOvCg8QtIefVLiW0xTyjch6gutu2SJQcHGn506Te3EGpYLNhym1THBD HSe0T8pR/2zMte2bZdtS77zrH0n53nJYkOGxc6xesuvb5MJJt41+6+HsdJ3MVl97 8lPry00itzYbU0yIJySH7PU5DePyFIzHOHzk6xk6dV+uCJfXGd1WaqjNuRtpFbMH R4/xAf2DPTxQ5N9QE0uWVOSdjbA8tGWG2zolY/fgAZu+qwMx5dM= =eIIz -----END PGP SIGNATURE-----