confile: pointer 'retv' is dereferenced.

[2xsec]

Hello,

If 'retv' pointer is null, snprintf can derefernce a null pointer.
snprintf allows null pointer with length = 0. In case of this, calling snprintf and other codes are inefficient.

Signed-off-by: Donghwa Jeong dh48.jeong@samsung.com

[lxc-jenkins]

This pull request didn't trigger Jenkins as its author isn't in the whitelist.

An organization member must perform one of the following:

• To have this branch tested by Jenkins, use the "ok to test" command.
• To have a one time test done, use the "test this please" command.

Those commands are simple Github comments of the format: "jenkins: COMMAND"

[brauner]

jenkins: test this please

[brauner]

The idea is that users can pass in NULL for retv in which case we report back the number of bytes that need to be allocated for the config item in question to be retrieved, i.e.:

	/*!
	 * \brief Retrieve the value of a config item.
	 *
	 * \param c Container.
	 * \param key Name of option to get.
	 * \param[out] retv Caller-allocated buffer to write value of \p key
	 * into (or \c NULL to determine length of value).
	 * \param inlen Length of \p retv (may be zero).
	 *
	 * \return Length of config items value, or < 0 on error.
	 *
	 * \note The caller can (and should) determine how large a buffer to allocate for
	 *  \p retv by initially passing its value as \c NULL and considering the return value.
	 *  This function can then be called again passing a newly-allocated suitably-sized buffer.
	 * \note If \p retv is NULL, \p inlen is ignored.
	 * \note If \p inlen is smaller than required, nothing will be written to \p retv and still return
	 *  the length of config item value.
	 */
	int (*get_config_item)(struct lxc_container *c, const char *key, char *retv, int inlen);

Your change would break this use-case. :)

[2xsec]

Yes, you are right. However, As far as I know, C99 allows str to be null but SUSv2 is not.
Anyway, thanks for your review.

[brauner]

@2xsec, ah, I see what you're getting at here. Let me think for a sec.

[brauner]

Actually, SUSv2 specifies a return value less than 1 so it should be safe but we can improve this a little.

[brauner]

SUSv2 shouldn't be in use anymore but we can make it safe by passing in a non-null buffer but 0 as size.

[2xsec]

OK, I'll try to make it.

[brauner]

Oh, sorry I jumped the gun in #2365. :) I wait for your comment next time. :)

[2xsec]

That's OK^^

[brauner]

Closed in favor of #2365. Thanks!