Check for potential null pointer

[SafeCoding233]

apr_palloc may return NULL pointer. It may trigger undefined behavior in later usage. For example, passing a null pointer to memcpy may trigger undefined behavior. Thus it is better to check for NULL.

[notroj]

With APR pools, the accepted coding style is to assume apr_palloc() never fails - applications can and install an abort function using apr_pool_create_ex() which is invoked if the underlying memory allocation ever fails.

[SafeCoding233]

With APR pools, the accepted coding style is to assume apr_palloc() never fails - applications can and install an abort function using apr_pool_create_ex() which is invoked if the underlying memory allocation ever fails.

Thank you for your kind reply and explanation!
I also thought the assumption is apr_palloc() never fails. However, I made the PR because I found some code snippets in the repo that indeed check for the possible null pointers.

For example:

apr/threadproc/netware/threadpriv.c

Lines 25 to 27 in 4283c6a

	(*key) = (apr_threadkey_t *)apr_palloc(pool, sizeof(apr_threadkey_t));
	if ((*key) == NULL) {
	return APR_ENOMEM;

apr/jose/apr_jose.c

Lines 71 to 73 in 4283c6a

	jose->jose.json = apr_palloc(pool, sizeof(apr_jose_json_t));
	if (!jose->jose.json) {
	return NULL;

So I thought it might be good to follow a consistent assumption?

I must admit that I didn't have experience contributing to the APR repo before. My apologies if my understanding is inaccurate, and thank you again for your time!

[notroj]

You're right that not all APR code follows the convention, unfortunately.