On Mon, 22 Feb 2010, Wolfram Sang wrote:
> Hello,
>
> I have a semantic patch which basically works, but does not produce
> spaces/blank lines as I'd like them. The spatch:
>
> ===
>
> @ init @
> type T;
> identifier client, data;
> @@
> T data = i2c_get_clientdata(client);
>
> @ main @
> type init.T;
> identifier init.client, init.data;
> statement S,S1;
> @@
>
> ... when != S
> + T data = i2c_get_clientdata(client);
> +
> S1
> ...
> + i2c_set_clientdata(client, NULL);
> + kfree(data);
> - kfree(i2c_get_clientdata(client));
>
> ===
>
> produces for the kernel-legacy-eeprom-driver (drivers/misc/eeprom):
>
> ===
>
> init_defs_builtins: /usr/local/share/coccinelle/standard.h
> HANDLING: eeprom.c
> diff =
> --- eeprom.c 2010-02-21 02:12:36.000000000 +0100
> +++ /tmp/cocci-output-17399-5d204d-eeprom.c 2010-02-22 08:54:46.000000000
> +0100
> @@ -208,8 +208,10 @@ exit:
>
> static int eeprom_remove(struct i2c_client *client)
> {
> + struct eeprom_data * data = i2c_get_clientdata(client);
> sysfs_remove_bin_file(&client->dev.kobj, &eeprom_attr);
> - kfree(i2c_get_clientdata(client));
> + i2c_set_clientdata(client, NULL);
> + kfree(data);
>
> return 0;
> }
>
> ===
>
> Note the missing empty line after the first added line. Also, there is an
> unwanted space after the asterisk. How do I fix those issues?
I can try to fix the space after the * issue.
The missing blank line, however, is more problematic, because there is no
representation for a blank line in the SmPL abstract syntax. It might be
possible to add that as well, though.
> (Side-question: I use V0.2.0. Couldn't I do something in the main-rule like
> 'extends init'? I just get parsing errors when I try to do it)
This is supposed to work. I will see what happens.
julia
_______________________________________________
Cocci mailing list
[email protected]
http://lists.diku.dk/mailman/listinfo/cocci
(Web access from inside DIKUs LAN only)
