6.6-stable review patch. If anyone has any objections, please let me know. ------------------ From: Breno Leitao [ Upstream commit edf9088b6e1d6d88982db7eb5e736a0e4fbcc09e ] Now that all other accesses to curr_xfer are done under the lock, protect the curr_xfer NULL check in tegra_qspi_isr_thread() with the spinlock. Without this protection, the following race can occur: CPU0 (ISR thread) CPU1 (timeout path) ---------------- ------------------- if (!tqspi->curr_xfer) // sees non-NULL spin_lock() tqspi->curr_xfer = NULL spin_unlock() handle_*_xfer() spin_lock() t = tqspi->curr_xfer // NULL! ... t->len ... // NULL dereference! With this patch, all curr_xfer accesses are now properly synchronized. Although all accesses to curr_xfer are done under the lock, in tegra_qspi_isr_thread() it checks for NULL, releases the lock and reacquires it later in handle_cpu_based_xfer()/handle_dma_based_xfer(). There is a potential for an update in between, which could cause a NULL pointer dereference. To handle this, add a NULL check inside the handlers after acquiring the lock. This ensures that if the timeout path has already cleared curr_xfer, the handler will safely return without dereferencing the NULL pointer. Fixes: b4e002d8a7ce ("spi: tegra210-quad: Fix timeout handling") Signed-off-by: Breno Leitao Tested-by: Jon Hunter Acked-by: Jon Hunter Acked-by: Thierry Reding Link: https://patch.msgid.link/20260126-tegra_xfer-v2-6-6d2115e4f387@debian.org Signed-off-by: Mark Brown Signed-off-by: Sasha Levin --- drivers/spi/spi-tegra210-quad.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/drivers/spi/spi-tegra210-quad.c b/drivers/spi/spi-tegra210-quad.c index f7677b960c342..8af5892ac738e 100644 --- a/drivers/spi/spi-tegra210-quad.c +++ b/drivers/spi/spi-tegra210-quad.c @@ -1351,6 +1351,11 @@ static irqreturn_t handle_cpu_based_xfer(struct tegra_qspi *tqspi) spin_lock_irqsave(&tqspi->lock, flags); t = tqspi->curr_xfer; + if (!t) { + spin_unlock_irqrestore(&tqspi->lock, flags); + return IRQ_HANDLED; + } + if (tqspi->tx_status || tqspi->rx_status) { tegra_qspi_handle_error(tqspi); complete(&tqspi->xfer_completion); @@ -1419,6 +1424,11 @@ static irqreturn_t handle_dma_based_xfer(struct tegra_qspi *tqspi) spin_lock_irqsave(&tqspi->lock, flags); t = tqspi->curr_xfer; + if (!t) { + spin_unlock_irqrestore(&tqspi->lock, flags); + return IRQ_HANDLED; + } + if (err) { tegra_qspi_dma_unmap_xfer(tqspi, t); tegra_qspi_handle_error(tqspi); @@ -1457,6 +1467,7 @@ static irqreturn_t handle_dma_based_xfer(struct tegra_qspi *tqspi) static irqreturn_t tegra_qspi_isr_thread(int irq, void *context_data) { struct tegra_qspi *tqspi = context_data; + unsigned long flags; u32 status; /* @@ -1474,7 +1485,9 @@ static irqreturn_t tegra_qspi_isr_thread(int irq, void *context_data) * If no transfer is in progress, check if this was a real interrupt * that the timeout handler already processed, or a spurious one. */ + spin_lock_irqsave(&tqspi->lock, flags); if (!tqspi->curr_xfer) { + spin_unlock_irqrestore(&tqspi->lock, flags); /* Spurious interrupt - transfer not ready */ if (!(status & QSPI_RDY)) return IRQ_NONE; @@ -1491,7 +1504,14 @@ static irqreturn_t tegra_qspi_isr_thread(int irq, void *context_data) tqspi->rx_status = tqspi->status_reg & (QSPI_RX_FIFO_OVF | QSPI_RX_FIFO_UNF); tegra_qspi_mask_clear_irq(tqspi); + spin_unlock_irqrestore(&tqspi->lock, flags); + /* + * Lock is released here but handlers safely re-check curr_xfer under + * lock before dereferencing. + * DMA handler also needs to sleep in wait_for_completion_*(), which + * cannot be done while holding spinlock. + */ if (!tqspi->is_curr_dma_xfer) return handle_cpu_based_xfer(tqspi); -- 2.53.0