fix(c9xx): don't flush dcache when invalidating

The data cache invalidation function for c9xx CPUs uses `dcache.cipa`
instruction. According to T-Head extension specification[1] section
3.1.5, this instruction also performs a cache clean along with the
invalidation.

On top of being incorrect, this leads to a serious issue on the
designware ethernet driver, where stalled cache may get flushed each
time we handle a new received packet[2]. As a result, received packet
are randomly corrupted with old cached data. This can easily be
reproduced by sending an ARP request to the device during a TFTP
transfer. The last TFTP block is treated as the ARP reply we just sent,
which makes U-Boot hang on the block.

Always use `dcache.ipa` instruction to invalidate dcache. Replace
existing usages of `dcache.ipa` with our implementation.

Note that this fix is slightly intrusive as it changes the cache
invalidation behavior in all drivers. However, I have not noticed any
side-effect during my tests.

[1] https://github.com/T-head-Semi/thead-extension-spec/releases/download/2.3.0/xthead-2023-11-10-2.3.0.pdf

[2] https://github.com/revyos/thead-u-boot/blob/918a8c89e056e3462031d6a498bb4fcc0c3526ce/drivers/net/designware.c#L475

5 files changed, 3 insertions, 17 deletions

diff --git a/arch/riscv/cpu/c9xx/cpu.c b/arch/riscv/cpu/c9xx/cpu.c
index e5eaed94..a93a3060 100644
--- a/arch/riscv/cpu/c9xx/cpu.c
+++ b/arch/riscv/cpu/c9xx/cpu.c
@@ -105,16 +105,6 @@ void invalidate_dcache_range(unsigned long start, unsigned long end)
 	register unsigned long i asm("a0") = start & ~(CONFIG_SYS_CACHELINE_SIZE - 1);
 
 	for (; i < end; i += CONFIG_SYS_CACHELINE_SIZE)
-		asm volatile(".long 0x02b5000b");  /* dcache.cipa a0 */
-
-	sync_is();
-}
-
-void invalid_dcache_range(unsigned long start, unsigned long end)
-{
-	register unsigned long i asm("a0") = start & ~(CONFIG_SYS_CACHELINE_SIZE - 1);
-
-	for (; i < end; i += CONFIG_SYS_CACHELINE_SIZE)
 		asm volatile(".long 0x02a5000b");  /* dcache.ipa a0 */
 
 	sync_is();
diff --git a/cmd/ddrscan.c b/cmd/ddrscan.c
index 0f2b78c6..c550e03d 100644
--- a/cmd/ddrscan.c
+++ b/cmd/ddrscan.c
@@ -73,7 +73,6 @@ extern ulong mem_test_quick(vu_long *buf, ulong start_addr, ulong end_addr,
 #endif
 extern void flush_dcache_range(unsigned long start, unsigned long end);
 extern void invalidate_dcache_range(unsigned long start, unsigned long end);
-extern void invalid_dcache_range(unsigned long start, unsigned long end);
 
 #ifdef CONFIG_CMD_MEMTEST
 int test_stuck_address(ulv *bufa, ulong count);
diff --git a/cmd/prbs.c b/cmd/prbs.c
index 2bbb68cf..59471169 100644
--- a/cmd/prbs.c
+++ b/cmd/prbs.c
@@ -50,7 +50,6 @@ u64 t_end;
 
 extern void flush_dcache_range(unsigned long start, unsigned long end);
 extern void invalidate_dcache_range(unsigned long start, unsigned long end);
-extern void invalid_dcache_range(unsigned long start, unsigned long end);
 
 extern unsigned long get_ddr_density(void);
 extern int riscv_get_time(u64 *time);
@@ -305,7 +304,7 @@ int prbs_test(struct PRBS_ELE *prbs, unsigned int *buf, int pos, bool random_dq,
     // compare result
     // invalid cache before read
     mdelay(100);
-    invalid_dcache_range((ulong)buf, (ulong)buf+(bit_len*4*2*2));
+    invalidate_dcache_range((ulong)buf, (ulong)buf+(bit_len*4*2*2));
     p1 = buf;
     bit_cnt = 0;
     for (i = 0; i < bit_len; i++) {
diff --git a/drivers/mmc/sdhci.c b/drivers/mmc/sdhci.c
index 5cc70cda..349e2bc1 100644
--- a/drivers/mmc/sdhci.c
+++ b/drivers/mmc/sdhci.c
@@ -247,10 +247,9 @@ static int sdhci_transfer_data(struct sdhci_host *host, struct mmc_data *data)
 		}
 	} while (!(stat & SDHCI_INT_DATA_END));
 #ifdef CONFIG_TARGET_LIGHT_C910	
-	extern void invalid_dcache_range(unsigned long start, unsigned long end);
 	/*After read ,invalid dcache range again to avoid cache filled during read tranfer*/
 	if(data->flags == MMC_DATA_READ){
-		invalid_dcache_range(host->start_addr,host->start_addr+ROUND(data->blocks*data->blocksize, ARCH_DMA_MINALIGN));
+		invalidate_dcache_range(host->start_addr,host->start_addr+ROUND(data->blocks*data->blocksize, ARCH_DMA_MINALIGN));
 	}
 #endif
 	return 0;
diff --git a/drivers/usb/dwc3/ep0.c b/drivers/usb/dwc3/ep0.c
index ea21f36d..1e07bdf8 100644
--- a/drivers/usb/dwc3/ep0.c
+++ b/drivers/usb/dwc3/ep0.c
@@ -906,8 +906,7 @@ static void dwc3_ep0_xfer_complete(struct dwc3 *dwc,
 	dep->resource_index = 0;
 	dwc->setup_packet_pending = false;
 #ifdef CONFIG_TARGET_LIGHT_C910
-	extern void invalid_dcache_range(unsigned long start, unsigned long end);
-	invalid_dcache_range((unsigned long)dwc->ctrl_req, (dmaaddr_t)dwc->ctrl_req + ROUND(sizeof(*dwc->ctrl_req), CACHELINE_SIZE));
+	invalidate_dcache_range((unsigned long)dwc->ctrl_req, (dmaaddr_t)dwc->ctrl_req + ROUND(sizeof(*dwc->ctrl_req), CACHELINE_SIZE));
 #endif
 
 	switch (dwc->ep0state) {