Feature/remove sed runtime dependency #683
Conversation
| bool copy_r = make_copy(RESOLV_CONF_FILE, RESOLV_CONF_FILE ".bkp"); | ||
|
|
||
| const char *match = "nameserver "; | ||
| off_t replace_size = strlen(match) + strlen(addr) + sizeof(char); |
There was a problem hiding this comment.
We should add 2 * sizeof(char) since since the replace string now includes a newline (in addition to the terminating nul).
There was a problem hiding this comment.
This is handled in #L466. Accounting for the null byte is always handled in the allocation functions in this code so that I could keep it consistent across malloc and realloc when computing the realloc size in #L533. In other words, *_size here indicates the size without the null bytes.
| strcpy(replace, match); | ||
| strcat(replace, addr); | ||
| strcat(replace, "\n"); |
There was a problem hiding this comment.
strncat would help to prevent overruns here.
building the replace string with a single call to snprintf might be cleaner.
| } | ||
| remaining_content[remaining_size] = '\0'; | ||
|
|
||
| _cleanup_(cleanup_bufferp) char *rptr = realloc(replace, (size_t)(replace_size + remaining_size + 1)); |
There was a problem hiding this comment.
why use a new variable here? I thought the common realloc pattern was p = realloc(p, size); ?
There was a problem hiding this comment.
I had some doubts about correct usage here after careful reading of the man page (it has very quirky behavior), so I took inspiration from information shared here:
https://stackoverflow.com/questions/44789295/correct-use-of-realloc
Additionally, there is interplay with the cleanup() function guard here I'm being careful to consider so we don't double free() in the case where the allocated memory block is moved from it's original location since realloc() will free() it internally. See #L534-544
No description provided.