[PATCH v2] Cygwin: clipboard: Fix a bug in read().

[Thomas Wolff]

Am 08.12.2021 um 11:19 schrieb Corinna Vinschen:
> On Dec  8 01:43, Mark Geisert wrote:
>> Takashi Yano wrote:
>> [...]
>>> I think the following patch makes the intent clearer.
>>> What do you think?
>>>
>>>
>>>   From d0aee9af225384a24ac6301f987ce2e94f262500 Mon Sep 17 00:00:00 2001
>>> From: Takashi Yano <takashi.yano@nifty.ne.jp>
>>> Date: Wed, 8 Dec 2021 17:06:03 +0900
>>> Subject: [PATCH] Cygwin: clipboard: Make intent of the code clearer.
>>>
>>> ---
>>>    winsup/cygwin/fhandler_clipboard.cc   | 4 ++--
>>>    winsup/cygwin/include/sys/clipboard.h | 1 +
>>>    2 files changed, 3 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/winsup/cygwin/fhandler_clipboard.cc b/winsup/cygwin/fhandler_clipboard.cc
>>> index 05f54ffb3..65a3cad97 100644
>>> --- a/winsup/cygwin/fhandler_clipboard.cc
>>> +++ b/winsup/cygwin/fhandler_clipboard.cc
>>> @@ -76,7 +76,7 @@ fhandler_dev_clipboard::set_clipboard (const void *buf, size_t len)
>>>          clipbuf->cb_sec  = clipbuf->ts.tv_sec;
>>>    #endif
>>>          clipbuf->cb_size = len;
>>> -      memcpy (&clipbuf[1], buf, len); // append user-supplied data
>>> +      memcpy (clipbuf->data, buf, len); // append user-supplied data
>>>          GlobalUnlock (hmem);
>>>          EmptyClipboard ();
>>> @@ -229,7 +229,7 @@ fhandler_dev_clipboard::read (void *ptr, size_t& len)
>>>          if (pos < (off_t) clipbuf->cb_size)
>>>    	{
>>>    	  ret = (len > (clipbuf->cb_size - pos)) ? clipbuf->cb_size - pos : len;
>>> -	  memcpy (ptr, (char *) (clipbuf + 1) + pos, ret);
>>> +	  memcpy (ptr, clipbuf->data + pos, ret);
>>>    	  pos += ret;
>>>    	}
>>>        }
>>> diff --git a/winsup/cygwin/include/sys/clipboard.h b/winsup/cygwin/include/sys/clipboard.h
>>> index 4c00c8ea1..b2544be85 100644
>>> --- a/winsup/cygwin/include/sys/clipboard.h
>>> +++ b/winsup/cygwin/include/sys/clipboard.h
>>> @@ -44,6 +44,7 @@ typedef struct
>>>        };
>>>      };
>>>      uint64_t      cb_size; // 8 bytes everywhere
>>> +  char          data[];
>>>    } cygcb_t;
>>>    #endif
>> Sigh.  I guess it's not possible to keep rid of a data item like I'd hoped.
>> At least "data[]" is cleaner than the historical "data[1]" here.  If you
>> call the item cb_data I can live with it.
>> Thanks all for the discussion.
>    sometype *ptr;
>
>    ptr = (sometype *) somebuffer;
>    do_something (ptr + 1);
>
> is a perfectly valid and perfectly readable thing, and used a lot if
> "sometype" is either a header in a buffer followed by arbitrary data, or
> if the buffer consists of multiple packed blocks of type "sometype".
>
> Takashi's suggestion adds the information that "sometype" is a header
> followed by arbitrary data, so that's a good thing..
Yes, thanks for this variant.
Thomas