[PATCH v2] ash: reorder control characters to save bytes

[Denys Vlasenko]

Let's keep the same values or at least the order as in dash.

There were bugs related to handling incoming chars which match "escape chars".
For example:

shell/ash_test/ash-quoting/dollar_squote_bash2.tests
shell/ash_test/ash-quoting/unicode_8x_chars.tests
shell/ash_test/ash-redir/redir7.tests
shell/ash_test/ash-redir/redir8.tests

If we change values of "escape chars", dash and ash will start exhibiting
*different behavior* if there are still bugs in this area =>
more PITA when debugging.

On Fri, Nov 16, 2018 at 2:48 PM Ron Yorston <rmy at pobox.com> wrote:
>
> During the processing of argument strings ash uses non-ASCII values
> to indicate particular constructs.  These values are arbitrary and
> can be changed at will.
>
> Sampling a number of permutations showed that the current choice
> of values results in the largest code.  11% of samples were just
> as bad and 89% were better.  The average reduction in code size
> was 11.77 bytes.
>
> function                                             old     new   delta
> cmdputs                                              393     383     -10
> argstr                                              1313    1301     -12
> .rodata                                           168610  168594     -16
> ------------------------------------------------------------------------------
> (add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-38)             Total: -38 bytes
>
> v2: For completeness also update comments in syntax_index_table.
>
> Signed-off-by: Ron Yorston <rmy at pobox.com>
> ---
>  shell/ash.c | 30 +++++++++++++++---------------
>  1 file changed, 15 insertions(+), 15 deletions(-)
>
> diff --git a/shell/ash.c b/shell/ash.c
> index 90eaf6faf..7ae81ec7a 100644
> --- a/shell/ash.c
> +++ b/shell/ash.c
> @@ -715,15 +715,15 @@ out2str(const char *p)
>  /* ============ Parser structures */
>
>  /* control characters in argument strings */
> -#define CTL_FIRST CTLESC
> -#define CTLESC       ((unsigned char)'\201')    /* escape next character */
> +#define CTL_FIRST CTLENDARI
> +#define CTLENDARI    ((unsigned char)'\201')
>  #define CTLVAR       ((unsigned char)'\202')    /* variable defn */
> -#define CTLENDVAR    ((unsigned char)'\203')
> -#define CTLBACKQ     ((unsigned char)'\204')
> -#define CTLARI       ((unsigned char)'\206')    /* arithmetic expression */
> -#define CTLENDARI    ((unsigned char)'\207')
> -#define CTLQUOTEMARK ((unsigned char)'\210')
> -#define CTL_LAST CTLQUOTEMARK
> +#define CTLESC       ((unsigned char)'\203')    /* escape next character */
> +#define CTLQUOTEMARK ((unsigned char)'\204')
> +#define CTLENDVAR    ((unsigned char)'\205')
> +#define CTLBACKQ     ((unsigned char)'\206')
> +#define CTLARI       ((unsigned char)'\207')    /* arithmetic expression */
> +#define CTL_LAST CTLARI
>
>  /* variable substitution byte (follows CTLVAR) */
>  #define VSTYPE  0x0f            /* type of variable substitution */
> @@ -3156,14 +3156,14 @@ static const uint8_t syntax_index_table[] ALIGN1 = {
>         /* 126  "~" */ CWORD_CCTL_CCTL_CWORD,
>         /* 127  del */ CWORD_CWORD_CWORD_CWORD,
>         /* 128 0x80 */ CWORD_CWORD_CWORD_CWORD,
> -       /* 129 CTLESC       */ CCTL_CCTL_CCTL_CCTL,
> +       /* 129 CTLENDARI    */ CCTL_CCTL_CCTL_CCTL,
>         /* 130 CTLVAR       */ CCTL_CCTL_CCTL_CCTL,
> -       /* 131 CTLENDVAR    */ CCTL_CCTL_CCTL_CCTL,
> -       /* 132 CTLBACKQ     */ CCTL_CCTL_CCTL_CCTL,
> -       /* 133 CTLQUOTE     */ CCTL_CCTL_CCTL_CCTL,
> -       /* 134 CTLARI       */ CCTL_CCTL_CCTL_CCTL,
> -       /* 135 CTLENDARI    */ CCTL_CCTL_CCTL_CCTL,
> -       /* 136 CTLQUOTEMARK */ CCTL_CCTL_CCTL_CCTL,
> +       /* 131 CTLESC       */ CCTL_CCTL_CCTL_CCTL,
> +       /* 132 CTLQUOTEMARK */ CCTL_CCTL_CCTL_CCTL,
> +       /* 133 CTLENDVAR    */ CCTL_CCTL_CCTL_CCTL,
> +       /* 134 CTLBACKQ     */ CCTL_CCTL_CCTL_CCTL,
> +       /* 135 CTLARI       */ CCTL_CCTL_CCTL_CCTL,
> +       /* 136      */ CWORD_CWORD_CWORD_CWORD,
>         /* 137      */ CWORD_CWORD_CWORD_CWORD,
>         /* 138      */ CWORD_CWORD_CWORD_CWORD,
>         /* 139      */ CWORD_CWORD_CWORD_CWORD,
> --
> 2.19.1
>
> _______________________________________________
> busybox mailing list
> busybox at busybox.net
> http://lists.busybox.net/mailman/listinfo/busybox