/[packages]/cauldron/kernel/current/PATCHES/patches/0091-scsi-esp_scsi-Track-residual-for-PIO-transfers.patch
ViewVC logotype

Contents of /cauldron/kernel/current/PATCHES/patches/0091-scsi-esp_scsi-Track-residual-for-PIO-transfers.patch

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1329221 - (show annotations) (download)
Fri Nov 9 22:05:45 2018 UTC (5 years, 9 months ago) by tmb
File size: 2426 byte(s)
add fixes from sashas autosel queue
1 From 0f201366b80bb62665636fab69f508842946161c Mon Sep 17 00:00:00 2001
2 From: Finn Thain <fthain@telegraphics.com.au>
3 Date: Tue, 16 Oct 2018 16:31:25 +1100
4 Subject: [PATCH 091/145] scsi: esp_scsi: Track residual for PIO transfers
5
6 [ Upstream commit fd47d919d0c336e7c22862b51ee94927ffea227a ]
7
8 If a target disconnects during a PIO data transfer the command may fail
9 when the target reconnects:
10
11 scsi host1: DMA length is zero!
12 scsi host1: cur adr[04380000] len[00000000]
13
14 The scsi bus is then reset. This happens because the residual reached
15 zero before the transfer was completed.
16
17 The usual residual calculation relies on the Transfer Count registers.
18 That works for DMA transfers but not for PIO transfers. Fix the problem
19 by storing the PIO transfer residual and using that to correctly
20 calculate bytes_sent.
21
22 Fixes: 6fe07aaffbf0 ("[SCSI] m68k: new mac_esp scsi driver")
23 Tested-by: Stan Johnson <userm57@yahoo.com>
24 Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
25 Tested-by: Michael Schmitz <schmitzmic@gmail.com>
26 Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
27 Signed-off-by: Sasha Levin <sashal@kernel.org>
28 ---
29 drivers/scsi/esp_scsi.c | 1 +
30 drivers/scsi/esp_scsi.h | 2 ++
31 drivers/scsi/mac_esp.c | 2 ++
32 3 files changed, 5 insertions(+)
33
34 diff --git a/drivers/scsi/esp_scsi.c b/drivers/scsi/esp_scsi.c
35 index c3fc34b9964d..9e5d3f7d29ae 100644
36 --- a/drivers/scsi/esp_scsi.c
37 +++ b/drivers/scsi/esp_scsi.c
38 @@ -1338,6 +1338,7 @@ static int esp_data_bytes_sent(struct esp *esp, struct esp_cmd_entry *ent,
39
40 bytes_sent = esp->data_dma_len;
41 bytes_sent -= ecount;
42 + bytes_sent -= esp->send_cmd_residual;
43
44 /*
45 * The am53c974 has a DMA 'pecularity'. The doc states:
46 diff --git a/drivers/scsi/esp_scsi.h b/drivers/scsi/esp_scsi.h
47 index 8163dca2071b..a77772777a30 100644
48 --- a/drivers/scsi/esp_scsi.h
49 +++ b/drivers/scsi/esp_scsi.h
50 @@ -540,6 +540,8 @@ struct esp {
51
52 void *dma;
53 int dmarev;
54 +
55 + u32 send_cmd_residual;
56 };
57
58 /* A front-end driver for the ESP chip should do the following in
59 diff --git a/drivers/scsi/mac_esp.c b/drivers/scsi/mac_esp.c
60 index eb551f3cc471..71879f2207e0 100644
61 --- a/drivers/scsi/mac_esp.c
62 +++ b/drivers/scsi/mac_esp.c
63 @@ -427,6 +427,8 @@ static void mac_esp_send_pio_cmd(struct esp *esp, u32 addr, u32 esp_count,
64 scsi_esp_cmd(esp, ESP_CMD_TI);
65 }
66 }
67 +
68 + esp->send_cmd_residual = esp_count;
69 }
70
71 static int mac_esp_irq_pending(struct esp *esp)
72 --
73 2.19.1
74

  ViewVC Help
Powered by ViewVC 1.1.30