[U-Boot] [PATCH v2] drivers: usb: fsl-dt-fixup: Fix the dt for multiple USB nodes in single traversal of device tree
Marek Vasut
marex at denx.de
Fri Sep 16 11:17:52 CEST 2016
On 09/16/2016 10:47 AM, Sriram Dash wrote:
>> From: Marek Vasut [mailto:marex at denx.de]
>> On 09/15/2016 12:29 AM, york sun wrote:
>>> On 09/14/2016 02:35 PM, Marek Vasut wrote:
>>>> On 09/14/2016 07:22 AM, Sriram Dash wrote:
>>>>>> From: Sriram Dash [mailto:sriram.dash at nxp.com]
>>>>>>
>>>>>
>>>>> Hello Marek,
>>>>>
>>>>> Any comments?
>>>>
>>>> Waiting for York to review this.
>>>>
>>>> It's a bulky patch V2 without changelog, shall I review the whole
>>>> thing again ?
>>>>
>>>>>> For FSL USB node fixup, the dt is walked multiple times for fixing
>>>>>> erratum and phy type. This patch walks the tree and fixes the node till no
>> more USB nodes are left.
>>>>>>
>>>>>> Signed-off-by: Sriram Dash <sriram.dash at nxp.com>
>>>>>> Signed-off-by: Rajesh Bhagat <rajesh.bhagat at nxp.com>
>>>>>> ---
>>>>>> drivers/usb/common/fsl-dt-fixup.c | 108
>>>>>> +++++++++++++++++---------------------
>>>>>> 1 file changed, 47 insertions(+), 61 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/usb/common/fsl-dt-fixup.c
>>>>>> b/drivers/usb/common/fsl-dt-fixup.c
>>>>>> index 9c48852..df785a6 100644
>>>>>> --- a/drivers/usb/common/fsl-dt-fixup.c
>>>>>> +++ b/drivers/usb/common/fsl-dt-fixup.c
>>>>>> @@ -54,25 +54,19 @@ static int fdt_usb_get_node_type(void *blob,
>>>>>> int start_offset, }
>>>>>>
>>>>>> static int fdt_fixup_usb_mode_phy_type(void *blob, const char *mode,
>>>>>> - const char *phy_type, int start_offset)
>>>>>> + const char *phy_type, int node_offset,
>>>>>> + const char **node_type)
>>>>>> {
>>>>>> const char *prop_mode = "dr_mode";
>>>>>> const char *prop_type = "phy_type";
>>>>>> - const char *node_type = NULL;
>>>>>> - int node_offset;
>>>>>> - int err;
>>>>>> -
>>>>>> - err = fdt_usb_get_node_type(blob, start_offset,
>>>>>> - &node_offset, &node_type);
>>>>>> - if (err < 0)
>>>>>> - return err;
>>>>>> + int err = 0;
>>>>>>
>>>>>> if (mode) {
>>>>>> err = fdt_setprop(blob, node_offset, prop_mode, mode,
>>>>>> strlen(mode) + 1);
>>>>>> if (err < 0)
>>>>>> printf("WARNING: could not set %s for %s: %s.\n",
>>>>>> - prop_mode, node_type, fdt_strerror(err));
>>>>>> + prop_mode, *node_type, fdt_strerror(err));
>>>>>> }
>>>>>>
>>>>>> if (phy_type) {
>>>>>> @@ -80,52 +74,48 @@ static int fdt_fixup_usb_mode_phy_type(void
>>>>>> *blob, const char *mode,
>>>>>> strlen(phy_type) + 1);
>>>>>> if (err < 0)
>>>>>> printf("WARNING: could not set %s for %s: %s.\n",
>>>>>> - prop_type, node_type, fdt_strerror(err));
>>>>>> + prop_type, *node_type, fdt_strerror(err));
>>>>>> }
>>>>>>
>>>>>> - return node_offset;
>>>>>> + return err;
>>>>>> }
>>>>>>
>>>>>> static int fdt_fixup_usb_erratum(void *blob, const char *prop_erratum,
>>>>>> - const char *controller_type, int start_offset)
>>>>>> + const char *controller_type, int node_offset,
>>>>>> + const char **node_type)
>>>>>> {
>>>>>> - int node_offset, err;
>>>>>> - const char *node_type = NULL;
>>>>>> + int err = -1;
>>>>>> const char *node_name = NULL;
>>>>>>
>>>>>> - err = fdt_usb_get_node_type(blob, start_offset,
>>>>>> - &node_offset, &node_type);
>>>>>> - if (err < 0)
>>>>>> - return err;
>>>>>> -
>>>>>> - if (!strcmp(node_type, FSL_USB2_MPH) || !strcmp(node_type,
>>>>>> FSL_USB2_DR))
>>>>>> + if (!strcmp(*node_type, FSL_USB2_MPH) ||
>>>>>> + !strcmp(*node_type, FSL_USB2_DR))
>>>>>> node_name = CHIPIDEA_USB2;
>>>>>> else
>>>>>> - node_name = node_type;
>>>>>> + node_name = *node_type;
>>>>>> if (strcmp(node_name, controller_type))
>>>>>> return err;
>>>>>>
>>>>>> err = fdt_setprop(blob, node_offset, prop_erratum, NULL, 0);
>>>>>> if (err < 0) {
>>>>>> printf("ERROR: could not set %s for %s: %s.\n",
>>>>>> - prop_erratum, node_type, fdt_strerror(err));
>>>>>> + prop_erratum, *node_type, fdt_strerror(err));
>>>>>> }
>>>>>>
>>>>>> - return node_offset;
>>>>>> + return err;
>>>>>> }
>>>>>>
>>>>>> -static int fdt_fixup_erratum(int *usb_erratum_off, void *blob,
>>>>>> +static int fdt_fixup_erratum(int node_offset, void *blob,
>>>>>> const char *controller_type, char *str,
>>>>>> - bool (*has_erratum)(void))
>>>>>> + bool (*has_erratum)(void), const char **node_type)
>>>>>> {
>>>>>> char buf[32] = {0};
>>>>>>
>>>>>> snprintf(buf, sizeof(buf), "fsl,usb-erratum-%s", str);
>>>>>> if (!has_erratum())
>>>>>> return -EINVAL;
>>>>>> - *usb_erratum_off = fdt_fixup_usb_erratum(blob, buf, controller_type,
>>>>>> - *usb_erratum_off);
>>>>>> - if (*usb_erratum_off < 0)
>>>>>> + node_offset = fdt_fixup_usb_erratum(blob, buf, controller_type,
>>>>>> + node_offset, node_type);
>>>>>> + if (node_offset < 0)
>>>>>> return -ENOSPC;
>>>>>> debug("Adding USB erratum %s\n", str);
>>>>>> return 0;
>>>>>> @@ -135,23 +125,23 @@ void fdt_fixup_dr_usb(void *blob, bd_t *bd) {
>>>>>> static const char * const modes[] = { "host", "peripheral", "otg" };
>>>>>> static const char * const phys[] = { "ulpi", "utmi", "utmi_dual" };
>>>>>> - int usb_erratum_a006261_off = -1;
>>>>>> - int usb_erratum_a007075_off = -1;
>>>>>> - int usb_erratum_a007792_off = -1;
>>>>>> - int usb_erratum_a005697_off = -1;
>>>>>> - int usb_erratum_a008751_off = -1;
>>>>>> - int usb_mode_off = -1;
>>>>>> - int usb_phy_off = -1;
>>>>>> + const char *node_type = NULL;
>>>>>> + int node_offset = -1;
>>>>>> char str[5];
>>>>>> - int i, j;
>>>>>> - int ret;
>>>>>> + int i = 1, j;
>>>>>> + int ret, err;
>>>>>>
>>>>>> - for (i = 1; i <= CONFIG_USB_MAX_CONTROLLER_COUNT; i++) {
>>>>>> + do {
>>>>>> const char *dr_mode_type = NULL;
>>>>>> const char *dr_phy_type = NULL;
>>>>>> int mode_idx = -1, phy_idx = -1;
>>>>>>
>>>>>> - snprintf(str, 5, "%s%d", "usb", i);
>>>>>> + err = fdt_usb_get_node_type(blob, node_offset,
>>>>>> + &node_offset, &node_type);
>>>>>> + if (err < 0)
>>>>>> + return;
>>>>>> +
>>>>>> + snprintf(str, 5, "%s%d", "usb", i++);
>>>>>> if (hwconfig(str)) {
>>>>>> for (j = 0; j < ARRAY_SIZE(modes); j++) {
>>>>>> if (hwconfig_subarg_cmp(str, "dr_mode", @@ -
>>>>>> 184,45 +174,41 @@ void fdt_fixup_dr_usb(void *blob, bd_t *bd)
>>>>>> if (has_dual_phy())
>>>>>> dr_phy_type = phys[2];
>>>>>>
>>>>>> - usb_mode_off = fdt_fixup_usb_mode_phy_type(blob,
>>>>>> - dr_mode_type, NULL,
>>>>>> - usb_mode_off);
>>>>>> -
>>>>>> - if (usb_mode_off < 0)
>>>>>> + err = fdt_fixup_usb_mode_phy_type(blob, dr_mode_type, NULL,
>>>>>> + node_offset, &node_type);
>>>>>> + if (err < 0)
>>>>>> return;
>>>>>>
>>>>>> - usb_phy_off = fdt_fixup_usb_mode_phy_type(blob,
>>>>>> - NULL, dr_phy_type,
>>>>>> - usb_phy_off);
>>>>>> -
>>>>>> - if (usb_phy_off < 0)
>>>>>> + err = fdt_fixup_usb_mode_phy_type(blob, NULL, dr_phy_type,
>>>>>> + node_offset, &node_type);
>>>>>> + if (err < 0)
>>>>>> return;
>>>>>>
>>>>>> - ret = fdt_fixup_erratum(&usb_erratum_a006261_off, blob,
>>>>>> + ret = fdt_fixup_erratum(node_offset, blob,
>>>>>> CHIPIDEA_USB2, "a006261",
>>>>>> - has_erratum_a006261);
>>>>>> + has_erratum_a006261, &node_type);
>>>>>> if (ret == -ENOSPC)
>>>>>> return;
>>>>>> - ret = fdt_fixup_erratum(&usb_erratum_a007075_off, blob,
>>>>>> + ret = fdt_fixup_erratum(node_offset, blob,
>>>>>> CHIPIDEA_USB2, "a007075",
>>>>>> - has_erratum_a007075);
>>>>>> + has_erratum_a007075, &node_type);
>>>>>> if (ret == -ENOSPC)
>>>>>> return;
>>>>>> - ret = fdt_fixup_erratum(&usb_erratum_a007792_off, blob,
>>>>>> + ret = fdt_fixup_erratum(node_offset, blob,
>>>>>> CHIPIDEA_USB2, "a007792",
>>>>>> - has_erratum_a007792);
>>>>>> + has_erratum_a007792, &node_type);
>>>>>> if (ret == -ENOSPC)
>>>>>> return;
>>>>>> - ret = fdt_fixup_erratum(&usb_erratum_a005697_off, blob,
>>>>>> + ret = fdt_fixup_erratum(node_offset, blob,
>>>>>> CHIPIDEA_USB2, "a005697",
>>>>>> - has_erratum_a005697);
>>>>>> + has_erratum_a005697, &node_type);
>>>>>> if (ret == -ENOSPC)
>>>>>> return;
>>>>>> - ret = fdt_fixup_erratum(&usb_erratum_a008751_off, blob,
>>>>>> + ret = fdt_fixup_erratum(node_offset, blob,
>>>>>> SNPS_DWC3, "a008751",
>>>>>> - has_erratum_a008751);
>>>>>> + has_erratum_a008751, &node_type);
>>>>>> if (ret == -ENOSPC)
>>>>>> return;
>>>>>>
>>>
>>> Do we really want to return in case of each error? I am not USB expert
>>> so my comment could be mistaken.
>>
>> The fdt_fixup_erratum() function is named in the most confusing dumb way, it is
>> not a generic function but a FSL/NXP specific one. Except it does look like a generic
>> one. This should've never made it mainline and it should be renamed.
>>
>
> Yes. I agree to your point. We will rename the functions and send a patch
> differently for the same.
Good
>> Regarding the return above, it's just that whoever implemented that function did it
>> in the most broken way possible. ENOSPC, according to
>> errno(3) means "ENOSPC No space left on device (POSIX.1)", which is
>> completely unrelated to anything USB in this context. Worse yet, the error is
>
> Yes. The ENOSPC is not related to USB. Also, this code fixes the device tree, which
> will not affect usb functionality in U boot.
OK
>> returned whenever fdt_fixup_usb_erratum() returns negative value , instead of
>> propagating errors.
>>
>
> The error is returned when fixup_usb_erratum returns a negative value.
> Now there are two possibilities.
> - The erratum is not to be applicable for the Soc, where the fdt_fixup_erratum
> returns EINVAL
> if (!has_erratum())
> return -EINVAL;
> In this case, we want to continue applying other errata.
> - The erratum is to be applicable, but the fdt_fixup_usb_erratum fails to apply it.
> if (node_offset < 0)
> return -ENOSPC;
> In this case, the failure will be inability to apply the erratum due to lack of
> space in device tree. So, check for the ENOSPC and return from the function
> fdt_fixup_dr_usb, as no other erratum can be applied further.
And what happens if fdt_fixup_usb_erratum() doesn't fail because of not
enough space in DT ? Then the information gets lost in here. So I don't
really care about this elaborate explanation, this should be fixed and
the errors should be propagated correctly.
> As York also pointed out regarding return in case of each error, we can avoid the
> returns and this will make the code clean and simple. But, the fdt_fixup_dr_usb
> will get executed till the device tree is traversed completely, regardless of the
> modifications taking effect over device tree.
> What do you think?
I think you should fix your error paths before you tackle this error
handling here at all, otherwise it's gonna be an annoying mess.
The way I'd do this is I'd have a table of fdt_fixup_erratum() arguments
and just iterate over the table. Then the code becomes
minimal.
>> I think the aforementioned two points should be fixed first and only then should the
>> return value above be handled correctly based on what the value is.
>>
>> I am real disappointed when looking at this crap.
>>
>>> Other than that, this patch looks OK. I didn't test it by compiling or
>>> running on a board.
>>
>> Please do.
>>
>>> York
>>>
>>
>>
>> --
>> Best regards,
>> Marek Vasut
--
Best regards,
Marek Vasut
More information about the U-Boot
mailing list