Subject: [GATOS]Trying to get TV-Out working for a Radeon 7000 w/ Xv support - msg#00007

Previous Message by Date: click to view message preview

[GATOS]Gibabyte GV-R92127VH VIVO is now working with ati.4.4.0!

Hello, All Well, my Gigabyte GV-R92128VH VIVO card (RV280 chip,RT100 Rage Theater chip) is now working. At first, I get ati.4.4.0 source code from CVS. apply two patches. After this i get km source code from Download section (km from CVS is unusable). Applyed patch #3, compileed all and now i can see movie in xawtv window. P.S. Trouble was in function RADEONI2CWriteRead and R200WriteRead (patch #2). I hope my patches will be helpful for the GATOS project. P.P.S. I can't get Aview working. I tested aview 0.12.7,aview 0.12.6 and aview 0.11.12 All were hanged up after starting and xawtv does not show video after aview startup (need reboot). So i cant't capture video. Regards, Vladimir Voroshilov 1. This patch is need for km to work (initialization of radeon memory controller). I havn't any docs, so code was copy/pasted from ati.2 source and tested only on my card. I only changed info->agpSize to info->gartSize diff -ur ati.4.4.0/radeon_driver.c ati.4.4.0.bld/radeon_driver.c --- ati.4.4.0/radeon_driver.c 2004-05-19 21:37:59.000000000 -0400 +++ ati.4.4.0.bld/radeon_driver.c 2004-09-05 18:55:26.000000000 -0400 @@ -2227,8 +2227,38 @@ } else + { pScrn->videoRam = INREG(RADEON_CONFIG_MEMSIZE) / 1024; + /* + This is initializing of radeon memory controler . + It is required for km capture module. + (copy/paste from ati.2 source code) + Tested only on RV280 chip + */ + CARD32 mc_fb_location; + CARD32 mc_agp_location; + CARD32 membase; + CARD32 memsize; + + memsize=INREG(RADEON_CONFIG_MEMSIZE); + membase = info->PciInfo->memBase[0] & 0xfc000000; + mc_fb_location = ((membase+memsize-1)&0xffff0000)|((membase>>16) & 0xffff); +#ifdef XF86DRI + mc_agp_location = ((membase+memsize+info->gartSize*1024*1024-1)&0xffff0000)|((membase+memsize)>>16); +#else + /* it is not being used - just hardcode value from BIOS */ + mc_agp_location = ((membase+memsize+0x100000-1)&0xffff0000)|((membase+memsize)>>16); + mc_agp_location = 0x003fffc0; +#endif + OUTREG(RADEON_MC_FB_LOCATION,mc_fb_location); + OUTREG(RADEON_MC_AGP_LOCATION,mc_agp_location); + OUTREG(RADEON_DISPLAY_BASE_ADDR,membase); + OUTREG(RADEON_OV0_BASE_ADDR,membase); + xf86DrvMsg(pScrn->scrnIndex, X_INFO, + "vid mem: membase=0x%x, memsize=%d, mc_fb_location=0x%x, mc_agp_location=0x%x\n", + membase,memsize,mc_fb_location,mc_agp_location); + } /* Some production boards of m6 will return 0 if it's 8 MB */ if (pScrn->videoRam == 0) pScrn->videoRam = 8192; 2. This patch is need for some card which havn't neither fi1236 nor saa7114 or msp2420 chips. I think that infinite loops is not a good idea. My card does not anwser on I2C bus (i don't no why) so foolowing code will couse infinitive lloop. while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); Some delays (like in X.org 6.7.0 I2C code) will be better, but i have not enough knowldge to write them. "For" loops couse some (about 20-40 secs) delays during X startup. Another temporary solution is removing fi1235,saa7114 and msp2420 files from multimedia directory. iff -ur ati.4.4.0/atixv.c ati.4.4.0.bld/atixv.c --- ati.4.4.0/atixv.c 2004-05-19 21:37:59.000000000 -0400 +++ ati.4.4.0.bld/atixv.c 2004-09-05 18:42:08.000000000 -0400 @@ -1455,7 +1455,7 @@ pATI->adaptor = adapt; - REGION_INIT(pScreen, &(pPriv->clip), NullBox, 0); + REGION_NULL(pScreen, &(pPriv->clip)); ATIResetVideo(pScrn); diff -ur ati.4.4.0/radeon_video.c ati.4.4.0.bld/radeon_video.c --- ati.4.4.0/radeon_video.c 2004-07-14 01:38:24.000000000 -0400 +++ ati.4.4.0.bld/radeon_video.c 2004-09-05 18:45:51.000000000 -0400 @@ -1010,7 +1010,9 @@ I2C_GO | I2C_START | ((nRead >0)?0:I2C_STOP) | I2C_DRIVE_EN; OUTREG(RADEON_I2C_CNTL_0, i2c_cntl_0); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -1038,7 +1040,9 @@ OUTREG(RADEON_I2C_CNTL_0, i2c_cntl_0); RADEONWaitForIdleMMIO(pScrn); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -1101,7 +1105,9 @@ write_mem_barrier(); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -1129,7 +1135,9 @@ OUTREG(RADEON_I2C_CNTL_0, i2c_cntl_0); write_mem_barrier(); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -2026,6 +2034,7 @@ pPriv = (RADEONPortPrivPtr)(adapt->pPortPrivates[0].ptr); /* XXX REGION_INIT(pScreen, &(pPriv->clip), NullBox, 0); */ + REGION_NULL(pScreen,&(pPriv->clip)); RADEONResetVideo(pScrn); if(info->VBIOS){ patch #3. It was need for correct detection of my card diff -ur km/km.c km.bld/km.c --- km/km.c 2003-12-14 14:38:37.000000000 -0500 +++ km.bld/km.c 2004-08-30 20:48:31.000000000 -0400 @@ -802,6 +802,10 @@ #define PCI_DEVICE_ID_ATI_RADEON_RD 0x5147 #endif +#ifndef PCI_DEVICE_ID_ATI_RADEON_Ya +#define PCI_DEVICE_ID_ATI_RADEON_Ya 0x5961 +#endif + static struct pci_device_id km_pci_tbl[] __devinitdata = { /* mach64 cards */ {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_215GB, @@ -954,6 +958,8 @@ PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RADEON_RD, PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, + {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RADEON_Ya, + PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, /* Radeon200 i.e. 8500 */ {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RADEON_BB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click

Next Message by Date: click to view message preview

[GATOS]new member/GATOS development plans

I just got write access to X.org CVS :)) Since I spent last few years maintaining GATOS project (http://gatos.sf.net), my first interest is to merge in existing stable code - after the HEAD is unfrozen. This will likely happen in the following order: 1. Radeon TV-in and miscellaneous I2C drivers 2. Rage128 TV-in 3. Mach64 TV-in Also, separately, I would like to merge CP engine code from http://volodya-project.sf.net/R300.php - this would require coordination with DRI and Mesa folks. best Vladimir Dergachev ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click

Previous Message by Thread: click to view message preview

[GATOS]Gibabyte GV-R92127VH VIVO is now working with ati.4.4.0!

Hello, All Well, my Gigabyte GV-R92128VH VIVO card (RV280 chip,RT100 Rage Theater chip) is now working. At first, I get ati.4.4.0 source code from CVS. apply two patches. After this i get km source code from Download section (km from CVS is unusable). Applyed patch #3, compileed all and now i can see movie in xawtv window. P.S. Trouble was in function RADEONI2CWriteRead and R200WriteRead (patch #2). I hope my patches will be helpful for the GATOS project. P.P.S. I can't get Aview working. I tested aview 0.12.7,aview 0.12.6 and aview 0.11.12 All were hanged up after starting and xawtv does not show video after aview startup (need reboot). So i cant't capture video. Regards, Vladimir Voroshilov 1. This patch is need for km to work (initialization of radeon memory controller). I havn't any docs, so code was copy/pasted from ati.2 source and tested only on my card. I only changed info->agpSize to info->gartSize diff -ur ati.4.4.0/radeon_driver.c ati.4.4.0.bld/radeon_driver.c --- ati.4.4.0/radeon_driver.c 2004-05-19 21:37:59.000000000 -0400 +++ ati.4.4.0.bld/radeon_driver.c 2004-09-05 18:55:26.000000000 -0400 @@ -2227,8 +2227,38 @@ } else + { pScrn->videoRam = INREG(RADEON_CONFIG_MEMSIZE) / 1024; + /* + This is initializing of radeon memory controler . + It is required for km capture module. + (copy/paste from ati.2 source code) + Tested only on RV280 chip + */ + CARD32 mc_fb_location; + CARD32 mc_agp_location; + CARD32 membase; + CARD32 memsize; + + memsize=INREG(RADEON_CONFIG_MEMSIZE); + membase = info->PciInfo->memBase[0] & 0xfc000000; + mc_fb_location = ((membase+memsize-1)&0xffff0000)|((membase>>16) & 0xffff); +#ifdef XF86DRI + mc_agp_location = ((membase+memsize+info->gartSize*1024*1024-1)&0xffff0000)|((membase+memsize)>>16); +#else + /* it is not being used - just hardcode value from BIOS */ + mc_agp_location = ((membase+memsize+0x100000-1)&0xffff0000)|((membase+memsize)>>16); + mc_agp_location = 0x003fffc0; +#endif + OUTREG(RADEON_MC_FB_LOCATION,mc_fb_location); + OUTREG(RADEON_MC_AGP_LOCATION,mc_agp_location); + OUTREG(RADEON_DISPLAY_BASE_ADDR,membase); + OUTREG(RADEON_OV0_BASE_ADDR,membase); + xf86DrvMsg(pScrn->scrnIndex, X_INFO, + "vid mem: membase=0x%x, memsize=%d, mc_fb_location=0x%x, mc_agp_location=0x%x\n", + membase,memsize,mc_fb_location,mc_agp_location); + } /* Some production boards of m6 will return 0 if it's 8 MB */ if (pScrn->videoRam == 0) pScrn->videoRam = 8192; 2. This patch is need for some card which havn't neither fi1236 nor saa7114 or msp2420 chips. I think that infinite loops is not a good idea. My card does not anwser on I2C bus (i don't no why) so foolowing code will couse infinitive lloop. while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); Some delays (like in X.org 6.7.0 I2C code) will be better, but i have not enough knowldge to write them. "For" loops couse some (about 20-40 secs) delays during X startup. Another temporary solution is removing fi1235,saa7114 and msp2420 files from multimedia directory. iff -ur ati.4.4.0/atixv.c ati.4.4.0.bld/atixv.c --- ati.4.4.0/atixv.c 2004-05-19 21:37:59.000000000 -0400 +++ ati.4.4.0.bld/atixv.c 2004-09-05 18:42:08.000000000 -0400 @@ -1455,7 +1455,7 @@ pATI->adaptor = adapt; - REGION_INIT(pScreen, &(pPriv->clip), NullBox, 0); + REGION_NULL(pScreen, &(pPriv->clip)); ATIResetVideo(pScrn); diff -ur ati.4.4.0/radeon_video.c ati.4.4.0.bld/radeon_video.c --- ati.4.4.0/radeon_video.c 2004-07-14 01:38:24.000000000 -0400 +++ ati.4.4.0.bld/radeon_video.c 2004-09-05 18:45:51.000000000 -0400 @@ -1010,7 +1010,9 @@ I2C_GO | I2C_START | ((nRead >0)?0:I2C_STOP) | I2C_DRIVE_EN; OUTREG(RADEON_I2C_CNTL_0, i2c_cntl_0); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -1038,7 +1040,9 @@ OUTREG(RADEON_I2C_CNTL_0, i2c_cntl_0); RADEONWaitForIdleMMIO(pScrn); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -1101,7 +1105,9 @@ write_mem_barrier(); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -1129,7 +1135,9 @@ OUTREG(RADEON_I2C_CNTL_0, i2c_cntl_0); write_mem_barrier(); - while(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8)); + for(loop=0;loop<1000000; loop++) + if (!(INREG8(RADEON_I2C_CNTL_0+1) & (I2C_GO >> 8))) + break; status=RADEON_I2C_WaitForAck(pScrn,pPriv); @@ -2026,6 +2034,7 @@ pPriv = (RADEONPortPrivPtr)(adapt->pPortPrivates[0].ptr); /* XXX REGION_INIT(pScreen, &(pPriv->clip), NullBox, 0); */ + REGION_NULL(pScreen,&(pPriv->clip)); RADEONResetVideo(pScrn); if(info->VBIOS){ patch #3. It was need for correct detection of my card diff -ur km/km.c km.bld/km.c --- km/km.c 2003-12-14 14:38:37.000000000 -0500 +++ km.bld/km.c 2004-08-30 20:48:31.000000000 -0400 @@ -802,6 +802,10 @@ #define PCI_DEVICE_ID_ATI_RADEON_RD 0x5147 #endif +#ifndef PCI_DEVICE_ID_ATI_RADEON_Ya +#define PCI_DEVICE_ID_ATI_RADEON_Ya 0x5961 +#endif + static struct pci_device_id km_pci_tbl[] __devinitdata = { /* mach64 cards */ {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_215GB, @@ -954,6 +958,8 @@ PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RADEON_RD, PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, + {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RADEON_Ya, + PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, /* Radeon200 i.e. 8500 */ {PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RADEON_BB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, HARDWARE_RADEON}, ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click

Next Message by Thread: click to view message preview

[GATOS]new member/GATOS development plans

I just got write access to X.org CVS :)) Since I spent last few years maintaining GATOS project (http://gatos.sf.net), my first interest is to merge in existing stable code - after the HEAD is unfrozen. This will likely happen in the following order: 1. Radeon TV-in and miscellaneous I2C drivers 2. Rage128 TV-in 3. Mach64 TV-in Also, separately, I would like to merge CP engine code from http://volodya-project.sf.net/R300.php - this would require coordination with DRI and Mesa folks. best Vladimir Dergachev ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click