From adf2a9a267422a2914a50b4a4a35a0e19b25d1c3 Mon Sep 17 00:00:00 2001 From: David Brownell Date: Tue, 5 Jan 2010 12:55:46 -0800 Subject: [PATCH] ARM: add comments re DAP assumptions I think some of these assumptions are not well-founded. Related, that swjdp_transaction_endcheck() is a bit iffy. Signed-off-by: David Brownell --- src/target/arm_adi_v5.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/src/target/arm_adi_v5.c b/src/target/arm_adi_v5.c index 82a2a28a..9f260642 100644 --- a/src/target/arm_adi_v5.c +++ b/src/target/arm_adi_v5.c @@ -195,8 +195,12 @@ static int scan_inout_check_u32(struct swjdp_common *swjdp, uint8_t instr, uint8_t reg_addr, uint8_t RnW, uint32_t outvalue, uint32_t *invalue) { + /* Issue the read or write */ adi_jtag_dp_scan_u32(swjdp, instr, reg_addr, RnW, outvalue, NULL, NULL); + /* For reads, collect posted value; RDBUFF has no other effect. + * Assumes read gets acked with OK/FAULT, and CTRL_STAT says "OK". + */ if ((RnW == DPAP_READ) && (invalue != NULL)) adi_jtag_dp_scan_u32(swjdp, JTAG_DP_DPACC, DP_RDBUFF, DPAP_READ, 0, invalue, &swjdp->ack); @@ -235,6 +239,9 @@ int swjdp_transaction_endcheck(struct swjdp_common *swjdp) /* Why??? second time it works??? */ #endif + /* Post CTRL/STAT read; discard any previous posted read value + * but collect its ACK status. + */ scan_inout_check_u32(swjdp, JTAG_DP_DPACC, DP_CTRL_STAT, DPAP_READ, 0, &ctrlstat); if ((retval = jtag_execute_queue()) != ERROR_OK) @@ -787,13 +794,18 @@ int mem_ap_read_buf_u32(struct swjdp_common *swjdp, uint8_t *buffer, int count, DPAP_READ, 0, NULL, NULL); for (readcount = 0; readcount < blocksize - 1; readcount++) { - /* Scan out read instruction and scan in previous value */ + /* Scan out next read; scan in posted value for the + * previous one. Assumes read is acked "OK/FAULT", + * and CTRL_STAT says that meant "OK". + */ adi_jtag_dp_scan(swjdp, JTAG_DP_APACC, AP_REG_DRW, DPAP_READ, 0, buffer + 4 * readcount, &swjdp->ack); } - /* Scan in last value */ + /* Scan in last posted value; RDBUFF has no other effect, + * assuming ack is OK/FAULT and CTRL_STAT says "OK". + */ adi_jtag_dp_scan(swjdp, JTAG_DP_DPACC, DP_RDBUFF, DPAP_READ, 0, buffer + 4 * readcount, &swjdp->ack); -- 2.39.5