ui: further reduce stack usage during mnemonic verification
What the developer wrote
Authored by Jon Griffiths
ui: further reduce stack usage during mnemonic verification
What changed, and why it matters
This commit is a small code cleanup in the part of Blockstream Jade that displays wallet backup words on screen. It changes how four word-pointers are passed between internal functions so they live in caller memory rather than being copied as function arguments. The stated goal is to reduce stack memory use during mnemonic verification. There is no direct evidence in the commit of a fix for an exploitable security bug, but reducing stack pressure can help prevent crashes or memory corruption in constrained hardware wallets.
Treat as a routine hardening/cleanup change. If auditing, verify the calling code still ensures all four word pointers are non-null and valid, and confirm the stack reduction is sufficient for the device's worst-case stack depth. No urgent action is indicated by this commit alone.
Security signals we found
Stack-usage reduction in a security-critical UI flow (mnemonic display)
Refactor only; no new bounds checks, sanitization, or cryptographic changes
No explicit vulnerability description or CVE reference in commit
Evidence from the diff
The patch refactors make_show_new_mnemonic_page() in main/ui/mnemonic.c to accept a single const char* words[4] array instead of four separate const char* parameters. The caller, make_show_mnemonic_activities(), now builds that array on its own stack before each call. This avoids pushing four pointer arguments per page and removes a local const char* words[] inside the callee, modestly lowering peak stack usage while displaying mnemonic words. No input validation, memory allocation, or cryptographic logic is changed.
Changed components
main/ui/mnemonic.cWallet mnemonic display/verification UIInspect captured patch +8 / −10
diff --git a/main/ui/mnemonic.c b/main/ui/mnemonic.c
index 7e37a8e..a50aded 100644
--- a/main/ui/mnemonic.c
+++ b/main/ui/mnemonic.c
@@ -104,14 +104,11 @@ gui_activity_t* make_bip85_mnemonic_words_activity(void)
return act;
}
-static void make_show_new_mnemonic_page(link_activity_t* page_act, const size_t nwords, const size_t first_index,
- const char* word1, const char* word2, const char* word3, const char* word4)
+static void make_show_new_mnemonic_page(
+ link_activity_t* page_act, const size_t nwords, const size_t first_index, const char* words[4])
{
JADE_ASSERT(page_act);
- JADE_ASSERT(word1);
- JADE_ASSERT(word2);
- JADE_ASSERT(word3);
- JADE_ASSERT(word4);
+ JADE_ASSERT(words && words[0] && words[1] && words[2] && words[3]);
// Support 12-word and 24-word mnemonics only
JADE_ASSERT(nwords == 12 || nwords == 24);
@@ -138,7 +135,6 @@ static void make_show_new_mnemonic_page(link_activity_t* page_act, const size_t
gui_set_parent(vsplit, parent);
char prefixed_word[16];
- const char* words[] = { word1, word2, word3, word4 };
for (int irow = 0; irow < 4; ++irow) {
// index-prefixed word, eg. "1: river"
const int ret = snprintf(prefixed_word, sizeof(prefixed_word), "%2u: %s", first_index + irow + 1, words[irow]);
@@ -174,10 +170,12 @@ void make_show_mnemonic_activities(gui_activity_t** first_activity_ptr, gui_acti
linked_activities_info_t act_info = {};
const size_t npages = nwords / 4; // 4 words per page
+ const char* words[4] = {};
for (size_t j = 0; j < npages; ++j) {
- uint16_t* offsets = word_offs + (j * 4);
- make_show_new_mnemonic_page(&page_act, nwords, j * 4, mnemonic + offsets[0], mnemonic + offsets[1],
- mnemonic + offsets[2], mnemonic + offsets[3]);
+ for (size_t w = 0; w < 4; ++w) {
+ words[w] = mnemonic + word_offs[j * 4 + w];
+ }
+ make_show_new_mnemonic_page(&page_act, nwords, j * 4, words);
gui_chain_activities(&page_act, &act_info);
}
Why this scored 27/100
Community notes
Notes can correct, qualify, or add evidence to the AI analysis. Every note shown here has been validated by a human moderator.
The AI analysis stands alone for now. Submit a note if you can add evidence or important context.