On Thu, 1 Oct 2026 18:46:00 +0530
Anshu Kumari
Implement option splitting per RFC 3396 for options that may exceed 255 bytes. A concat_req[] lookup table marks options requiring concatenation (currently option 81, Client FQDN per RFC 4702).
The opts[].s buffer is resized from 255 to OPT_CONCAT_MAX to hold the maximum data that can be split across the options field, file field, and sname field.
When a concatenation-requiring option does not fit as a single option in any field, fill() calls fill_split() to split it across fields in RFC 3396 order: options field first, then file, then sname. Each split chunk is capped at 255 bytes per the DHCP option length field limit.
Link: https://bugs.passt.top/show_bug.cgi?id=192 Signed-off-by: Anshu Kumari
--- v7: - Capped each split chunk at 255 bytes: chunk = MIN(MIN(remaining, avail), 255) to prevent overflow in the uint8_t length field - Replaced inline space calculation with a dry-run pass via fill_split_areas() - Moved OPT_MAX and OPT_CONCAT_MAX definitions above struct opt. - Reordered struct opt fields to place s[OPT_CONCAT_MAX] - Added dry_run parameter to fill_split() v6: - Merged v5 patches 6/7 and 7/7 into a single patch. - Replaced DHCP_OPT_STR_CONCAT enum value and is_concat_opt() helper with a concat_req[] boolean lookup table. - Used MIN() macro instead of ternary for chunk size. - Fixed space calculation to account for 2-byte code+length overhead per chunk.
v5: - New patch: implement option splitting per RFC 3396 for options exceeding 255 bytes - Add DHCP_OPT_STR_CONCAT type, is_concat_opt(), fill_split() helpers - Resize opts[].s from 255 to OPT_CONCAT_MAX (497) bytes - Add /* fallthrough */ between DHCP_OPT_STR and DHCP_OPT_STR_CONCAT case --- dhcp.c | 160 ++++++++++++++++++++++++++++++++++++++++++++++++++------- 1 file changed, 142 insertions(+), 18 deletions(-)
diff --git a/dhcp.c b/dhcp.c index 63ac8e3d..990b650d 100644 --- a/dhcp.c +++ b/dhcp.c @@ -35,6 +35,19 @@ #include "dhcp.h" #include "parse.h"
+#define OPT_MIN 60 /* RFC 951 */ + +/* Total option size (excluding end option) is 576 (RFC 2131), minus + * offset of options (268), minus end option (1). + */ +#define OPT_MAX 307 + +/* RFC 3396: maximum option data that can be split across options field + * (OPT_MAX - 2), file field (64 - 2), and sname field (128 - 2), + * minus code+length overhead per portion.
That doesn't match the define below (which is the correct one, I think). If you additionally subtract at the end, from all that, code and length bytes for each section, you would end up with an extra -6 term, which isn't needed.
+ */ +#define OPT_CONCAT_MAX (OPT_MAX - 2 + 64 - 2 + 128 - 2) + /** * enum opt_conf - DHCP option configuration * @OPT_UNSET: Option not configured @@ -51,18 +64,18 @@ enum opt_conf { * struct opt - DHCP option * @sent: Convenience flag, set while filling replies * @slen: Length of option defined for server - * @s: Option payload from server * @clen: Length of option received from client, -1 if not received * @c: Option payload from client * @conf: Option configuration (unset, default, or user) + * @s: Option payload from server */ struct opt { int sent; int slen; - uint8_t s[255]; int clen; uint8_t c[255]; enum opt_conf conf; + uint8_t s[OPT_CONCAT_MAX];
As I mentioned, clang-tidy (make clang-tidy) suggests rearranging this in a way that doesn't need padding.
};
static struct opt opts[256]; @@ -79,32 +92,23 @@ static struct opt opts[256]; #define DHCPINFORM 8 #define DHCPFORCERENEW 9
-#define OPT_MIN 60 /* RFC 951 */ - -/* Total option size (excluding end option) is 576 (RFC 2131), minus - * offset of options (268), minus end option (1). - */ -#define OPT_MAX 307 - /** * dhcp_init() - Initialise DHCP options */ void dhcp_init(void) { if (opts[1].conf != OPT_USER) /* Mask */ - opts[1] = (struct opt) { 0, 4, { 0 }, 0, { 0 }, OPT_DEFAULT }; + opts[1] = (struct opt) { 0, 4, 0, { 0 }, OPT_DEFAULT, { 0 } }; if (opts[3].conf != OPT_USER) /* Router */ - opts[3] = (struct opt) { 0, 4, { 0 }, 0, { 0 }, OPT_DEFAULT }; + opts[3] = (struct opt) { 0, 4, 0, { 0 }, OPT_DEFAULT, { 0 } }; if (opts[51].conf != OPT_USER) { /* Lease time */ - opts[51] = (struct opt) { 0, 4, { 0xff, 0xff, 0xff, 0xff }, - 0, { 0 }, OPT_DEFAULT }; + opts[51] = (struct opt) { 0, 4, 0, { 0 }, OPT_DEFAULT, + { 0xff, 0xff, 0xff, 0xff } }; } /* Type */ - opts[53] = (struct opt) { 0, 1, { 0 }, 0, { 0 }, OPT_DEFAULT }; - if (opts[54].conf != OPT_USER) { /* Server ID */ - opts[54] = (struct opt) { 0, 4, { 0 }, 0, { 0 }, - OPT_DEFAULT }; - } + opts[53] = (struct opt) { 0, 1, 0, { 0 }, OPT_DEFAULT, { 0 } }; + if (opts[54].conf != OPT_USER) /* Server ID */ + opts[54] = (struct opt) { 0, 4, 0, { 0 }, OPT_DEFAULT, { 0 } }; }
/** @@ -211,6 +215,11 @@ static const enum dhcp_opt_type dhcp_opt_types[] = { [252] = DHCP_OPT_STR, /* WPAD URL */ };
+/* Options requiring RFC 3396 concatenation, indexed by code */ +static const bool concat_req[256] = { + [81] = true, /* Client FQDN (RFC 4702, Section 2) */ +}; + /** * dhcp_opt_parse() - Parse a DHCP option value * @code: DHCP option code @@ -308,6 +317,9 @@ static int dhcp_opt_parse(uint8_t code, const char *str, case DHCP_OPT_STR: slen = strlen(str);
+ if (!concat_req[code] && slen > 255) + return -1; + if (slen > buf_len) return -1;
@@ -441,6 +453,45 @@ static void fill_one(uint8_t *buf, size_t size, int o, int *offset) *offset += slen; }
+/** + * fill_split() - Write a split portion of an option into a buffer + * @buf: Buffer to write into + * @size: Usable size of @buf + * @o: Option number (code) + * @offset: Current offset within @buf, updated on write + * @data: Pointer to remaining option data to write + * @remaining: Bytes of option data still to write + * @dry_run: If true, calculate size without writing to @buf + * + * Return: number of data bytes written (excluding code+length header) + */ +static size_t fill_split(uint8_t *buf, size_t size, int o, int *offset, + const uint8_t *data, size_t remaining, + bool dry_run) +{ + size_t avail, chunk; + + if (*offset + 2 >= (int)size) + return 0; + + avail = size - *offset - 2; + chunk = MIN(MIN(remaining, avail), 255); + if (!chunk) + return 0; + + if (!dry_run) { + buf[*offset] = o; + buf[*offset + 1] = chunk; + } + *offset += 2; + + if (!dry_run) + memcpy(buf + *offset, data, chunk); + *offset += chunk; + + return chunk; +} + /** * enum dhcp_overload - DHCP option overload values (RFC 2132, Section 9.3) * @DHCP_OVERLOAD_NONE: No overload @@ -453,6 +504,43 @@ enum dhcp_overload { DHCP_OVERLOAD_SNAME = 2, };
+/** + * fill_split_areas() - Write a split option across options, file, and sname areas + * @m: Message to write into + * @opt_size: Usable size of the options area + * @o: Option number (code) + * @opt_off: Current offset within options area, updated on write + * @file_off: Current offset within file area, updated on write + * @sname_off: Current offset within sname area, updated on write + * @has_bootfile: If true, skip the file field + * @dry_run: If true, calculate size without writing + * + * Return: total data bytes written across all areas + */ +static size_t fill_split_areas(struct msg *m, size_t opt_size, int o, + int *opt_off, int *file_off, int *sname_off, + bool has_bootfile, bool dry_run) +{ + size_t written = 0;
No need to initialise this to 0 and then increment it unconditionally below, I think it's a bit confusing.
+ + written += fill_split(m->o, opt_size, o, opt_off, + opts[o].s, opts[o].slen, dry_run); + if (written < (size_t)opts[o].slen && !has_bootfile) { + written += fill_split(m->file, sizeof(m->file) - 1, o, + file_off, + opts[o].s + written, + opts[o].slen - written, dry_run); + } + if (written < (size_t)opts[o].slen) { + written += fill_split(m->sname, sizeof(m->sname) - 1, o, + sname_off, + opts[o].s + written, + opts[o].slen - written, dry_run); + } + + return written; +} + /** * fill() - Fill options in message, with overload into file/sname if needed * @m: Message to fill @@ -504,6 +592,42 @@ static int fill(struct msg *m, bool has_bootfile) } }
+ /* RFC 3396: split concatenation-requiring options that didn't fit + * as a single option. Split order: options, file, sname. + */ + foreach_opt(o) {
This could also use the foreach_unsent_opt() iterator I was suggesting for 6/7.
+ int dry_off, dry_file_off, dry_sname_off; + size_t total, written; + + if (opts[o].conf == OPT_UNSET || opts[o].sent || + !concat_req[o])
+ continue; + + /* Dry run: verify the option fits across all areas */ + dry_off = offset; + dry_file_off = file_off; + dry_sname_off = sname_off; + + total = fill_split_areas(m, size, o, + &dry_off, &dry_file_off, + &dry_sname_off, + has_bootfile, true); + + if (total < (size_t)opts[o].slen) { + debug("DHCP: skipping option %i (no space to split)",
Splitting itself doesn't need extra space, it's simply that we don't have space for the option (not even if split). Sorry, I missed this on v6.
+ o); + continue; + } + + /* Actual write */ + written = fill_split_areas(m, size, o, + &offset, &file_off, &sname_off, + has_bootfile, false); + + if (written >= (size_t)opts[o].slen) + opts[o].sent = 1; + } + /* Report any options that could not be sent */ foreach_opt(o) { if (opts[o].conf != OPT_UNSET && !opts[o].sent)
-- Stefano