Re: [PATCH 1/6] lib/cmdline.c: Add backslash support to kernel commandline parsing.
From: Michal Suchánek <hidden>
Date: 2017-09-25 18:39:09
Also in:
lkml
On Fri, 15 Sep 2017 19:28:56 +0200 Michal Such=C3=A1nek [off-list ref] wrote:
On Fri, 15 Sep 2017 18:14:09 +0100 Al Viro [off-list ref] wrote: =20quoted
On Fri, Sep 15, 2017 at 07:02:46PM +0200, Michal Suchanek wrote: =20quoted
for (i =3D 0; args[i]; i++) { - if (isspace(args[i]) && !in_quote) + if (isspace(args[i]) && !in_quote && !backslash) break; - if (equals =3D=3D 0) { - if (args[i] =3D=3D '=3D') - equals =3D i; + + if ((equals =3D=3D 0) && (args[i] =3D=3D '=3D')) + equals =3D i; + + if (!backslash) { + if ((args[i] =3D=3D '"') || (args[i] =3D=3D '\\')) { + if (args[i] =3D=3D '"') + in_quote =3D !in_quote; + if (args[i] =3D=3D '\\') + backslash =3D 1; + + memmove(args + 1, args, i); + args++; + i--; + } + } else { + backslash =3D 0; } - if (args[i] =3D=3D '"') - in_quote =3D !in_quote; } =20=20 ... and that makes for Unidiomatic Work With Strings Award for this September. Using memmove() for string rewrite is almost always bad taste; in this case it's also (as usual) broken. =20=20 Care to share how it is broken?
Guess not. I will assume it is perfectly fine then. It works perfectly fine in my testing. Using memmove for string rewrite is not a matter of taste. It is the only library function with sane semantics for rewrite of anything. Then again open-coding it is always an option and maybe in better taste for some :-> Michal