mail archive of the barebox mailing list
 help / color / mirror / Atom feed
From: Sascha Hauer <s.hauer@pengutronix.de>
To: BAREBOX <barebox@lists.infradead.org>
Cc: Sascha Hauer <sascha@saschahauer.de>
Subject: [PATCH 3/6] partitions: gpt: validate mkpart range against GPT usable LBAs
Date: Wed, 01 Jul 2026 11:27:36 +0200	[thread overview]
Message-ID: <20260701-gpt-size-fixes-v1-3-aec2e5641a19@pengutronix.de> (raw)
In-Reply-To: <20260701-gpt-size-fixes-v1-0-aec2e5641a19@pengutronix.de>

efi_partition_mkpart() validated the requested partition range against
hardcoded device geometry: start_lba against a fixed 34 and end_lba
against last_lba(blk) - 33. These offsets assume the default layout of
128 partition entries (32 sectors) plus the backup GPT header.

The authoritative bounds are first_usable_lba and last_usable_lba stored
in the GPT header being modified. They can differ from the hardcoded
values, e.g. when a table is read from disk that was created for a
smaller device (last_usable_lba retains the smaller value until the GPT
is rewritten) or that reserves a differently sized partition entry
array. In those cases the old check allowed a partition to extend past
last_usable_lba, overlapping the reserved backup entry array and GPT
header.

Check start_lba/end_lba against the header's first_usable_lba and
last_usable_lba instead.

Signed-off-by: Sascha Hauer <sascha@saschahauer.de>
---
 common/partitions/efi.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/common/partitions/efi.c b/common/partitions/efi.c
index 98172672ff..59b9fa9b55 100644
--- a/common/partitions/efi.c
+++ b/common/partitions/efi.c
@@ -764,15 +764,18 @@ static __maybe_unused int efi_partition_mkpart(struct partition_desc *pd,
 	gpt_entry *pte;
 	int i;
 	const guid_t *guid;
+	uint64_t first_usable_lba = le64_to_cpu(gpt->first_usable_lba);
+	uint64_t last_usable_lba = le64_to_cpu(gpt->last_usable_lba);
 
-	if (start_lba < 34) {
-		pr_err("invalid start LBA %lld, minimum is 34\n", start_lba);
+	if (start_lba < first_usable_lba) {
+		pr_err("invalid start LBA %lld, minimum is %lld\n", start_lba,
+		       first_usable_lba);
 		return -EINVAL;
 	}
 
-	if (end_lba >= last_lba(pd->blk) - 33) {
+	if (end_lba > last_usable_lba) {
 		pr_err("invalid end LBA %lld, maximum is %lld\n", end_lba,
-		       last_lba(pd->blk) - 34);
+		       last_usable_lba);
 		return -EINVAL;
 	}
 

-- 
2.47.3




  parent reply	other threads:[~2026-07-01  9:28 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-07-01  9:27 [PATCH 0/6] GPT: fixes around last_usable_lba Sascha Hauer
2026-07-01  9:27 ` [PATCH 1/6] partitions: gpt: Adjust last_usable_lba to real device size when writing Sascha Hauer
2026-07-01  9:27 ` [PATCH 2/6] partitions: gpt: refresh also when primary or alternate is invalid Sascha Hauer
2026-07-01  9:27 ` Sascha Hauer [this message]
2026-07-01  9:27 ` [PATCH 4/6] partitions: check new partitions against usable LBAs, not device size Sascha Hauer
2026-07-01  9:27 ` [PATCH 5/6] partitions: gpt: fix wrong error messages Sascha Hauer
2026-07-01  9:27 ` [PATCH 6/6] partitions: gpt: log a message when partition table refreshed Sascha Hauer

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260701-gpt-size-fixes-v1-3-aec2e5641a19@pengutronix.de \
    --to=s.hauer@pengutronix.de \
    --cc=barebox@lists.infradead.org \
    --cc=sascha@saschahauer.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox