On 07.12.18 17:19, David Hildenbrand wrote:
> On 05.12.18 16:39, Janosch Frank wrote:
>> z/VM isn't fond of vt220, so we need line mode when running under z/VM.
>>
>> Signed-off-by: Janosch Frank <frankja@xxxxxxxxxxxxx>
>> ---
>> lib/s390x/sclp-console.c | 199 ++++++++++++++++++++++++++++++++++++++++++-----
>> lib/s390x/sclp.h | 67 ++++++++++++++++
>> 2 files changed, 245 insertions(+), 21 deletions(-)
>>
>> diff --git a/lib/s390x/sclp-console.c b/lib/s390x/sclp-console.c
>> index 13ab03d..c31e178 100644
>> --- a/lib/s390x/sclp-console.c
>> +++ b/lib/s390x/sclp-console.c
>> @@ -11,23 +11,183 @@
[...]
>> +
>> +static bool initialized = false;
>
> false is the default after your BSS clearing patch.
Yes, already fixed
>
>> +
>> +static void sclp_print_ascii(const char *str)
>> +{
>> + int len = strlen(str);
>> + WriteEventData *sccb = (void *)_sccb;
>> +
>> + while (sclp_busy)
>> + /* Wait for SCLP request to complete */;
>> + sclp_busy = true;
>> + memset(sccb, 0, sizeof(*sccb));
>> + sccb->h.length = sizeof(WriteEventData) + len;
>> + sccb->h.function_code = SCLP_FC_NORMAL_WRITE;
>> + sccb->ebh.length = sizeof(EventBufferHeader) + len;
>> + sccb->ebh.type = SCLP_EVENT_ASCII_CONSOLE_DATA;
>> + memcpy(sccb->data, str, len);
>> +
>> + sclp_service_call(SCLP_CMD_WRITE_EVENT_DATA, sccb);
>
> Shouldn't we wait for a response? (see below)
Yes, already fixed
>
>> +}
>> +
>> +static void sclp_print_lm(const char *str)
>> +{
>> + unsigned char *ptr, *end, ch;
>> + unsigned int count, offset, len;
>> + struct write_sccb *sccb;
>> + struct msg_buf *msg;
>> + struct mdb *mdb;
>> + struct mto *mto;
>> + struct go *go;
>> +
>> + while (sclp_busy)
>> + /* Wait for SCLP request to complete */;
>> + sclp_busy = true;
>> + sccb = (struct write_sccb *) _sccb;
>> + end = (unsigned char *) sccb + 4096 - 1;
>> + memset(sccb, 0, sizeof(*sccb));
>> + ptr = (unsigned char *) &sccb->msg.mdb.mto;
>> + len = strlen(str);
>> + offset = 0;
>> + do {
>> + for (count = sizeof(*mto); offset < len; count++) {
>> + ch = str[offset++];
>> + if ((ch == 0x0a) || (ptr + count > end))
>> + break;
>> + ptr[count] = _ascebc[ch];
>> + }
>> + mto = (struct mto *) ptr;
>> + memset(mto, 0, sizeof(*mto));
>> + mto->length = count;
>> + mto->type = 4;
>> + mto->line_type_flags = LNTPFLGS_ENDTEXT;
>> + ptr += count;
>> + } while ((offset < len) && (ptr + sizeof(*mto) <= end));
>> + len = ptr - (unsigned char *) sccb;
>> + sccb->header.length = len - offsetof(struct write_sccb, header);
>> + msg = &sccb->msg;
>> + msg->header.type = EVTYP_MSG;
>> + msg->header.length = len - offsetof(struct write_sccb, msg.header);
>> + mdb = &msg->mdb;
>> + mdb->header.type = 1;
>> + mdb->header.tag = 0xD4C4C240;
>> + mdb->header.revision_code = 1;
>> + mdb->header.length = len - offsetof(struct write_sccb, msg.mdb.header);
>> + go = &mdb->go;
>> + go->length = sizeof(*go);
>> + go->type = 1;
>> + sclp_service_call(SCLP_CMD_WRITE_EVENT_DATA, sccb);
>
> Shouldn't we wait for a response? (see below)
Yes, already fixed
>
>> +}
>> +
>> +/*
>> + * SCLP needs to be initialized by setting a send and receive mask,
>> + * indicating which messages the control program (we) want(s) to
>> + * send/receive.
>> + */
>> +static bool sclp_set_write_mask(void)
>> {
>> WriteEventMask *sccb = (void *)_sccb;
>>
>> while (sclp_busy)
>> /* Wait for SCLP request to complete */;
>
> Can this actually happen?
>
> I'd say whoever issues a request has to wait for the response. Somebody
> who didn't, doesn't have to wait.
We could put the sclp_wait_busy() at the end of the sclp call function
and effectively make it synchronous by enforcing the wait. We don't
really need the async performance gain.
>
> (for multiple CPUs later on that has to be guarded by a lock in addition
> to the bool)
Yeah, smp is currently on another branch, so I might just start to add
another patch and introduce locks. :-)
>
>> sclp_busy = true;
>> + memset(_sccb, 0, sizeof(*sccb));
>> sccb->h.length = sizeof(WriteEventMask);
>> - sccb->mask_length = sizeof(unsigned int);
>> - sccb->receive_mask = SCLP_EVENT_MASK_MSG_ASCII;
>> - sccb->cp_receive_mask = SCLP_EVENT_MASK_MSG_ASCII;
>> - sccb->send_mask = SCLP_EVENT_MASK_MSG_ASCII;
>> - sccb->cp_send_mask = SCLP_EVENT_MASK_MSG_ASCII;
>> + sccb->h.function_code = 0;
>> + sccb->mask_length = sizeof(sccb_mask_t);
>> +
>> + /* For now we don't process sclp input. */
>> + sccb->cp_receive_mask = 0;
>> + /* We send ASCII and line mode. */
>> + sccb->cp_send_mask = SCLP_EVENT_MASK_MSG_ASCII | SCLP_EVENT_MASK_MSG;
>>
>> sclp_service_call(SCLP_CMD_WRITE_EVENT_MASK, sccb);
>> + while (sclp_busy) {
>> + }
>> + if (sccb->h.response_code != SCLP_RC_NORMAL_COMPLETION)
>> + return false;
>
> Hmm, what do do in this case? I don't see that the caller is fixed up
> (you changed the return type from void to bool)
I don't know, there used to be a TODO there...
I thought about calling exit(1);
>
> I "initialized" actually worth keeping around? I'd say let's just
> continue as if it succeeded. (we cannot really do anything about it, can
> we?)
There is the problem of prints that come before the initialization.
PGM interrupts while sclp_busy == true is also fun if they want to print.
Attachment:
signature.asc
Description: OpenPGP digital signature
