On Sat, Nov 26, 2011 at 12:21 PM, Alexander Nezhinsky
<alexandern@xxxxxxxxxxxx> wrote:
> Text queries issued through tgtadm use a fixed-size buffer that is filled,
> step by step, typically by a few calls to snprintf() or an equivalent.
> This requires tracking the current position within the buffer and guarding
> against overflows. Currently these goals are achived by few separate macros:
> shprintf(), buffer_check() and _buffer_check().
> This approach relies upon local/global counter variables that describe
> the buffer state and in some cases a label to escape through on errors.
> Two kinds of problems arise in such settings:
> First, using hardcoded variable names and especially 'goto' statements in
> macros is unsafe, and second, it becomes difficult to use nested function
> calls if the called functions also append some strings to the buffer.
>
> The alternative offered by this patch uses a buffer descriptor context,
> 'struct concat_buf', which stores the buffer pointer and relevant counters
> and error code. This internal state information avoids writes when buffer is
> full and in all cases of previosuly encountered errors.
>
> Thus we can continue with the function flow as if the buffer is unlimited
> and in the end return an appropriate error code using concat_buf_ret().
> Code is simplified and the tedious checks afer each snprintf are avoided.
>
> As the context may be passed to other functions easily and its state is
> preserved upon return, nested calls are naturally enabled.
>
> Signed-off-by: Alexander Nezhinsky <alexandern@xxxxxxxxxxxx>
> ---
> usr/util.c | 26 ++++++++++++++++++++++++++
> usr/util.h | 40 +++++++++++++++++++++++++++++-----------
> 2 files changed, 55 insertions(+), 11 deletions(-)
>
> diff --git a/usr/util.c b/usr/util.c
> index c78a999..43d2690 100644
> --- a/usr/util.c
> +++ b/usr/util.c
> @@ -194,3 +194,29 @@ int get_blk_shift(unsigned int size)
> return shift;
> }
>
> +int concat_printf(struct concat_buf *b, const char *format, ...)
> +{
> + va_list args;
> + int rest = b->size - b->used;
> + int nprinted;
> +
> + if (b->err)
> + return b->err;
> + if (!rest)
> + return (b->err = -ENOSPC);
> +
> + va_start(args, format);
> + nprinted = vsnprintf(b->buf + b->used, rest, format, args);
> + va_end(args);
> +
> + if (nprinted < 0)
> + return (b->err = nprinted);
> +
> + if (nprinted >= rest) {
> + nprinted = rest;
> + b->err = -ENOSPC;
> + }
> + b->used += nprinted;
> + return b->err;
> +}
> +
> diff --git a/usr/util.h b/usr/util.h
> index 8abdb94..18d81b8 100644
> --- a/usr/util.h
> +++ b/usr/util.h
> @@ -8,6 +8,7 @@
> #include <signal.h>
> #include <syscall.h>
> #include <unistd.h>
> +#include <stdarg.h>
> #include <limits.h>
> #include <linux/types.h>
>
> @@ -88,17 +89,6 @@ static inline int between(uint32_t seq1, uint32_t seq2, uint32_t seq3)
> return seq3 - seq2 >= seq1 - seq2;
> }
>
> -#define shprintf(total, buf, rest, fmt, args...) \
> -do { \
> - int len; \
> - len = snprintf(buf, rest, fmt, ##args); \
> - if (len > rest) \
> - goto overflow; \
> - buf += len; \
> - total += len; \
> - rest -= len; \
> -} while (0)
> -
> extern unsigned long pagesize, pageshift;
>
> #if defined(__NR_signalfd) && defined(USE_SIGNALFD)
> @@ -151,4 +141,32 @@ struct signalfd_siginfo {
> ret; \
> })
>
> +struct concat_buf {
> + char *buf;
> + int size;
> + int used;
> + int err;
> +};
> +
> +static inline void concat_buf_init(struct concat_buf *b,
> + char *buf, int size)
> +{
> + b->buf = buf;
> + b->size = size;
> + b->used = 0;
> + b->err = 0;
> +}
> +
> +static inline int concat_buf_ret(struct concat_buf *b)
> +{
> + return !b->err ? b->used : b->err;
> +}
> +
> +static inline const char *concat_delim(struct concat_buf *b, const char *delim)
> +{
> + return !b->used ? "" : delim;
> +}
> +
> +extern int concat_printf(struct concat_buf *b, const char *format, ...);
> +
> #endif
Looks like a duplicate of asprintf() ? See also
http://www.gnu.org/s/libc/manual/html_node/Dynamic-Output.html.
Fubo.
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
