[U-Boot] [PATCH v3 01/13] efi_loader: support Unicode text input

[Alexander Graf]

On 11.09.18 22:38, Heinrich Schuchardt wrote:
> Up to now the EFI_TEXT_INPUT_PROTOCOL only supported ASCII characters.
> With the patch it can consume UTF-8 from the console.
> 
> Currently only the serial console and the console can deliver UTF-8.
> Local consoles are restricted to ASCII.
> 
> Signed-off-by: Heinrich Schuchardt <xypron.glpk at gmx.de>
> ---
> v2:
> 	drop support for German keyboard
> 	move reading of Unicode code to charset.c
> ---
>  include/charset.h            |   9 +++
>  lib/charset.c                | 136 ++++++++++++++++++++++-------------
>  lib/efi_loader/efi_console.c |  13 ++--
>  test/unicode_ut.c            |   8 +--
>  4 files changed, 108 insertions(+), 58 deletions(-)
> 
> diff --git a/include/charset.h b/include/charset.h
> index 686db5a1fe..a7de5f6948 100644
> --- a/include/charset.h
> +++ b/include/charset.h
> @@ -8,11 +8,20 @@
>  #ifndef __CHARSET_H_
>  #define __CHARSET_H_
>  
> +#include <efi.h>

Yeah ... eh ... no :).

I assume this is just a leftover from the old version?

>  #include <linux/kernel.h>
>  #include <linux/types.h>
>  
>  #define MAX_UTF8_PER_UTF16 3
>  
> +/**
> + * console_read_unicode() - read Unicode code point from console
> + *
> + * @code:	code point

Please specify this a bit clearer.

> + * Return:	0 = success
> + */
> +int console_read_unicode(s32 *code);
> +
>  /**
>   * utf8_get() - get next UTF-8 code point from buffer
>   *
> diff --git a/lib/charset.c b/lib/charset.c
> index 72c808ce64..1806b41cc3 100644
> --- a/lib/charset.c
> +++ b/lib/charset.c
> @@ -5,6 +5,7 @@
>   *  Copyright (c) 2017 Rob Clark
>   */
>  
> +#include <common.h>
>  #include <charset.h>
>  #include <capitalization.h>
>  #include <malloc.h>
> @@ -18,67 +19,106 @@ static struct capitalization_table capitalization_table[] =
>  	CP437_CAPITALIZATION_TABLE;
>  #endif
>  
> -s32 utf8_get(const char **src)
> +/**
> + * get_code() - read Unicode code point from UTF-8 stream
> + *
> + * @read_u8:	- stream reader
> + * @src:	- string buffer passed to stream reader, optional
> + * Return:	- Unicode code point
> + */
> +static int get_code(u8 (*read_u8)(void *data), void *data)
>  {
> -	s32 code = 0;
> -	unsigned char c;
> +	s32 ch = 0;
>  
> -	if (!src || !*src)
> -		return -1;
> -	if (!**src)
> +	ch = read_u8(data);
> +	if (!ch)
>  		return 0;
> -	c = **src;
> -	if (c >= 0x80) {
> -		++*src;
> -		if (!**src)
> -			return -1;
> -		/*
> -		 * We do not expect a continuation byte (0x80 - 0xbf).
> -		 * 0x80 is coded as 0xc2 0x80, so we cannot have less then 0xc2
> -		 * here.
> -		 * The highest code point is 0x10ffff which is coded as
> -		 * 0xf4 0x8f 0xbf 0xbf. So we cannot have a byte above 0xf4.
> -		 */
> -		if (c < 0xc2 || code > 0xf4)
> -			return -1;
> -		if (c >= 0xe0) {
> -			if (c >= 0xf0) {
> +	if (ch >= 0xc2 && ch <= 0xf4) {
> +		int code = 0;
> +
> +		if (ch >= 0xe0) {
> +			if (ch >= 0xf0) {
>  				/* 0xf0 - 0xf4 */
> -				c &= 0x07;
> -				code = c << 18;
> -				c = **src;
> -				++*src;
> -				if (!**src)
> -					return -1;
> -				if (c < 0x80 || c > 0xbf)
> -					return -1;
> -				c &= 0x3f;
> +				ch &= 0x07;
> +				code = ch << 18;
> +				ch = read_u8(data);
> +				if (ch < 0x80 || ch > 0xbf)
> +					goto error;
> +				ch &= 0x3f;
>  			} else {
>  				/* 0xe0 - 0xef */
> -				c &= 0x0f;
> +				ch &= 0x0f;
>  			}
> -			code += c << 12;
> +			code += ch << 12;
>  			if ((code >= 0xD800 && code <= 0xDFFF) ||
>  			    code >= 0x110000)
> -				return -1;
> -			c = **src;
> -			++*src;
> -			if (!**src)
> -				return -1;
> -			if (c < 0x80 || c > 0xbf)
> -				return -1;
> +				goto error;
> +			ch = read_u8(data);
> +			if (ch < 0x80 || ch > 0xbf)
> +				goto error;
>  		}
>  		/* 0xc0 - 0xdf or continuation byte (0x80 - 0xbf) */
> -		c &= 0x3f;
> -		code += c << 6;
> -		c = **src;
> -		if (c < 0x80 || c > 0xbf)
> -			return -1;
> -		c &= 0x3f;
> +		ch &= 0x3f;
> +		code += ch << 6;
> +		ch = read_u8(data);
> +		if (ch < 0x80 || ch > 0xbf)
> +			goto error;
> +		ch &= 0x3f;
> +		ch += code;
> +	} else if (ch >= 0x80) {
> +		goto error;
>  	}
> -	code += c;
> +	return ch;
> +error:
> +	return '?';
> +}
> +
> +/**
> + * read_string() - read byte from character string
> + *
> + * @data:	- pointer to string
> + * Return:	- byte read
> + *
> + * The string pointer is incremented if it does not point to '\0'.
> + */
> +static u8 read_string(void *data)
> +
> +{
> +	const char **src = (const char **)data;
> +	u8 c;
> +
> +	if (!src || !*src || !**src)
> +		return 0;
> +	c = (unsigned char)**src;

Please remove the cast. Btw, you could also write this as

  return *(*src++);

>  	++*src;
> -	return code;
> +	return c;
> +}
> +
> +/**
> + * read_console() - read byte from console
> + *
> + * @src		- not used, needed to match interface
> + * Return:	- byte read
> + */
> +static u8 read_console(void *data)
> +{
> +	return getc();
> +}
> +
> +int console_read_unicode(s32 *code)
> +{
> +	if (!tstc())
> +		/* No input available */
> +		return 1;

Please avoid multi-line indented code without braces.

> +
> +	/* Read Unicode code */
> +	*code = get_code(read_console, NULL);
> +	return 0;
> +}
> +
> +s32 utf8_get(const char **src)
> +{
> +	return get_code(read_string, src);
>  }
>  


Alex